Where did we get the idea that our entire nature, not just our embryological development, but everything else, was pre-programmed by our genome? After all, the very essence of Homo sapiens compared to all other species, is that we use culture--language, tools, etc.--to do our business rather than just our physical biology. In a serious sense, we evolved to be free of our bodies, our genes made us freer from our genes than most if not all other species! And we evolved to live long enough to learn--language, technology, etc.--in order to live our thus-long lives.
Yet isn't an assumption of pre-programming the only assumption by which anyone could legitimately promise 'precision' genomic medicine? Of course, Mendel's work, adopted by human geneticists over a century ago, allowed great progress in understanding how genes lead at least to the simpler of our traits, with discrete (yes/no) manifestations, traits that do include many diseases that really, perhaps surprisingly, do behave in Mendelian fashion, and for which concepts like dominance and recessiveness been applied and that, sometimes, at least approximately hold up to closer scrutiny.
Even 100 years ago, agricultural and other geneticists who could do experiments, largely confirmed the extension of Mendel to continuously varying traits, like blood pressure or height. They reasoned that many genes (whatever they were, which was unknown at the time) contributed individually small effects. If each gene had two states in the usual Aa/AA/aa classroom example sense, but there were countless such genes, their joint action could approximate continuously varying traits whose measure was, say, the number of A alleles in an individual. This view was also consistent with the observed correlation of trait measure with kinship-degree among relatives. This history has been thoroughly documented. But there are some bits, important bits, missing, especially when it comes to the fervor for Big Data 'omics analysis of human diseases and other traits. In essence, we are still, a century later, conceptual prisoners of Mendel.
'Omics over the top: key questions generally ignored
Let us take GWAS (genomewide association studies) on their face value. GWAS find countless 'hits', sites of whatever sort across the genome whose variation affects variation in WhateverTrait you choose to map (everything simply must be 'genomic' or some other 'omic, no?). WhateverTrait varies because every subject in your study has a different combination of contributing alleles. Somewhat resembling classical Mendelian recessiveness, contributing alleles are found in cases as well as controls (or across the measured range of quantitative traits like stature or blood pressure), where the measured trait reflects how many A's one has: WhateverTrait is essentially the sum of A's in 'cases', which may be interpreted as a risk--some sort of 'probability' rather than certainty--of having been affected or of having the measured trait value.
We usually treat risk as a 'probability,' a single value, p, that applies to everyone with the same genotype. Here, of course, no two subjects have exactly the same genotype so some sort of aggregate risk score, adding up each person's 'hits', is assigned a p. This, however, tacitly assumes something like that each site contributes some fixed risk or 'probability' of affection. But this treats these values as if they were essential to the site, each thus acting as a parameter of risk. That is, sites are treated as a kind of fixed value or, one might say 'force', relative to the trait measure in question.
One obvious and serious issue is that these are necessarily estimated from past data, that is, by induction from samples. Not only is there sampling variation that usually is only crudely estimated by some standard statistical variation-related measure, but we know that the picture will be at least somewhat different in any other sample we might have chosen, not to mention other populations; and those who are actually candid about what they are doing know very well that the same people living in a different place or time would have different risks for the same trait.
No study is perfect, so we use some conveniently assumed well-behaved regression/correction adjustments to account for the statistical 'noise' due to factors like age, sex, and unmeasured environmental effects. Much worse than these issues, there are clearly factors of imprecision, and the obvious major one, taboo even to think about much less to mention, that relevant future factors (mutations, environments, lifestyles) are unknowable, even in principle. So what we really do, are forced to do, is extend what the past was like to the assumed future. But besides this, we don't count somatic changes (mutation arising in body tissues during life, that were not inherited), because they'd mess up our assertions of 'precision', and we can't measure them well in any case (so just shut one's eyes and pretend the ghost isn't in the house!).
All of these together mean that we are estimating risks from imperfect existing samples and past life-experience, but treating them as underlying parameters so that we can extend them to future samples. What that does is equate induction with deduction, assuming the past is rigorously parametric and will be the same in the future; but this is simply scientifically and epistemologically wrong, no matter how inconvenient it is to acknowledge this. Mutations, genotypes, and environments of the future are simply unpredictable, even in principle.
None of this is a secret, or new discovery, in any way. What it is, is inconvenient truth. These things should have been enough, by themselves and without badgering investigators about environmental factors that (we know very well, typically predominate) prevent all the NIH's precision promises from being accurate ('precise'), or even to a knowable degree. Yet this 'precision' sloganeering is being, sheepishly, aped all over the country by all sorts of groups who don't think for themselves and/or who go along lest they get left off the funding gravy train. This is the 'omics fad. If you think I am being too cynical, just look at what's being said, done, published, and claimed.
These are, to me, deep flaws in the way the GWAS and other 'omics industries, very well-heeled, are operating these days, to pick the public's pocket (pharma may, slowly, be awakening-- Lancet editorial, "UK life science research: time to burst the biomedical bubble," Lancet 392:187, 2018). But scientists need jobs and salaries, and if we put people in a position where they have to sing in this way for their supper, what else can you expect of them?
Unfortunately, there are much more serious problems with the science, and they have to do with the point-cause thinking on which all of this is based.
Even a point-cause must act through some process
By far most of the traits, disease or otherwise, that are being GWAS'ed and 'omicked these days, at substantial public expense, are treated as if the mapped 'causes' are point causes. If there are n causes, and a person has an unlucky set m out of many possible sets, one adds 'em up and predicts that person will have the target trait. And there is much that is ignored, assumed, or wishfully hidden in this 'will'. It is not clear how many authors treat it, tacitly, as a probability vs a certainty, because no two people in a sample have the same genotype and all we know is that they are 'affected' or 'unaffected'.
The genomics industry promises, essentially, that from conception onward, your DNA sequence will predict your diseases, even if only in the form of some 'risk'; the latter is usually a probability and despite the guise of 'precision' it can, of course, be adjusted as we learn more. For example, it must be adjusted for age, and usually other variables. Thus, we need ever larger and more and longer-lasting samples. This alone should steer people away from being profiteered by DNA testing companies. But that snipe aside, what does this risk or 'probability' actually mean?
Among other things, those candid enough to admit it know that environmental and lifestyle factors have a role, interacting with the genotype if not, usually, overwhelming it, meaning, for example, that the genotype only confers some, often modest, risk probability, the actual risk much more affected by lifestyle factors, most of which are not measured or not measured with accuracy, or not even yet identified. And usually there is some aspect that relates to age, or some assumption about what 'lifetime' risk means. Whose lifetime?
Aspects of such a 'probability'
There are interesting issues, longstanding issues, about these probabilities, even if we assume they have some kind of meaning. Why do so many important diseases, like cancers, only arise at some advanced age? How can a genomic 'risk' be so delayed and so different among people? Why are mice, with very similar genotypes to humans (which is why we do experiments on them to learn about human disease) only live to 3 while we live to our 70s and beyond?
Richard Peto, raised some of these questions many decades ago. But they were never really addressed, even in an era when NIH et al were spending much money on 'aging' research including studies of lifespan. There were generic theories that suggested from an evolutionary theory why some diseases were deferred to later ages (it is called 'negative pleiotropy'), but nobody tried seriously to explain why that was from a molecular/genetic point of view. Why do mice only live only 3 years, anyway? And so on.
These are old questions and very deep ones but they have not been answered and, generally, are conveniently forgotten--because, one might argue, they are inconvenient.
If a GWAS score increases the risk of a disease, that has a long delayed onset pattern, often striking late in life, and highly variable among individuals or over time, what sort of 'cause' is that genotype? What is it that takes decades for the genes to affect the person? There are a number of plausible answers, but they get very little attention at least in part because that stands in the way of the vested interests of entrenched too-big-to-kill Big Data faddish 'research' that demands instant promises to the public it is trephining for support. If the major reason is lifestyle factors, then the very delayed onset should be taken as persuasive evidence that the genotype is, in fact, by itself not a very powerful predictor.
Why would the additive effects of some combination of GWAS hits lead to disease risk? That is, in our complex nature why would each gene's effects be independent of each other contributor? In fact, mapping studies usually show evidence that other things, such as interactions are important--but they are at present almost impossibly complex to be understood.
Does each combination of genome-wide variants have a separate age-onset pattern, and if not, why not? And if so, how does the age effect work (especially if not due to person-years of exposure to the truly determining factors of lifestyle)? If such factors are at play, how can we really know, since we never see the same genotype twice? How can we assume that the time-relationship with each suspect genetic variant will be similar among samples or in the future? Is the disease due to post-natal somatic mutation, in which case why make predictions based on the purported constitutive genotypes of GWAS samples?
Obviously, if long delayed onset patterns are due not to genetic but to lifestyle exposures interacting with genotypes, then perhaps lifestyle exposures should be the health-related target, not exotic genomic interventions. Of course, the value of genome-based prediction clearly depends on environmental/lifestyle exposures, and the future of these exposure is obviously unknowable (as we clearly do know from seeing how unpredictable past exposures have affected today's disease patterns).
The point here is that our reliance on genotypes is a very convenient way of keeping busy, bringing in the salaries, but not facing up to the much more challenging issues that the easy one (run lots of data through DNA sequencers) can't address. I did not invent these points, and it is hard to believe that at least the more capable and less me-too scientists don't clearly know them, if quietly. Indeed, I know this from direct experience. Yes, scientists are fallible, vain, and we're only human. But of all human endeavors, science should be based on honesty because we have to rely on trust of each other's work.
The scientific problems are profound and not easily solved, and not soluble in a hurry. But much of the problem comes from the funding and careerist system that shackles us. This is the deeper explanation in many ways. The paint on the House of Science is the science itself, but it is the House that supports that paint that is the real problem.
A civically responsible science community, and its governmental supporters, should be freed from the iron chains of relentless Big Data for their survival, and start thinking, seriously, about the questions that their very efforts over the past 20 years, on trait after trait, in population after population, and yes, with Big Data, have clearly revealed.
Showing posts with label complex disease. Show all posts
Showing posts with label complex disease. Show all posts
Tuesday, October 16, 2018
Friday, March 20, 2015
Science, causality and belief
By Anne Buchanan and Ken Weiss
Questions about cause and effect in the natural world have been bandied about since long before Aristotle's time. Aristotle, though, believed that the use of the idea of causality must be supported by theory. Without theory, he thought, a thorough and systematic investigation of the world, answering the 'why' questions, was not possible.
His theory, or categories of possible answers to why questions, consisted of these four causes (from the (Stanford Encyclopedia of Philosophy):
This week's episode of BBC Radio 4 program, In Our Time is a discussion of the life and thought of 12th century Islamic thinker Al-Ghazali. Al-Ghazali wrote about theology, jurisprudence, mysticism and philosophy, interesting in their own right, but what struck us was the discussion of his thinking about causality. His book, The Incoherence of the Philosophers, changed Islamic thinking about epistemology. Al-Ghazali vehemently rejected the material explanations of causation of Aristotle and others, instead proposing that all causality was due to the Will of God.
The example given on In Our Time was of cotton burning when it touches fire. Must the cotton burn? The philosophers who Al-Ghazali rejected would have said yes, but Al-Ghazali believed that the material world must be seen as based on God's relationship to the world. So, causality has to be viewed in relation to God, and from this perspective, the cotton could fail to burn because miracles are always possible. Or, because God makes everything happen, he could stop the burning. Or, yes, fire causes cotton to burn, but only because God intervenes in everything, and makes the cotton burn when it touches fire. It only looks as though there are physical causes to everything. In reality, God is the cause.
Of course, this view is easily dismissible by scientists, who believe in material causes. But science doesn't, in fact, always do much better. As was noted on In Our Time, David Hume, 18th century Scottish philosopher who played a central role in the Enlightenment, the approach to material questions about the world based on empirical, objective observation, had read Al-Ghazali and was influenced in part by what he read.
Hume would have said that, yes, we think that fire must burn cotton, but this is something we believe only out of habit. We've always seen cotton burn when it touches fire, but it's sloppy thinking to assume it will always burn. Indeed, he was a critic of this kind of inductive reasoning, the assumption that nature will always behave as we expect, based on our past observations, and will continue to be regular. The "all the swans we've seen are white, therefore all swans are white" pure induction kind of thinking. The problem, of course, is that the first black swan we see destroys our generality about swans, and brings us back to square one. Or, in practice, we try to cling to our theory and explain the exception (e.g., all white swans are white, the black one is a different species or type).
An alternative to induction is deduction -- "All men are mortal, Socrates is a man, Socrates is mortal." But, deduction relies on a starting generality that is assumed to be correct, and if it is not, the conclusion will be false as well. So, both forms of reasoning have fundamental problems, but both are still in use, in part for lack of better alternatives, and in part because in practice they are often useful. If gravity is universal and based on distance and mass, then we can predict the motion of objects we have never seen before, or we can understand the motion of multiple objects. We use deduction in genetics by reasoning from assumptions about what genes are and do to infer what genetically has caused some outcome (e.g., an instance of a disease). But here we too often don't have strong deductive rigor in our premises.
Since the 19th century and the birth of statistical approaches to evaluating the strength of association between risk factors and outcomes, though, in some ways science has sidestepped the problems with determining causality by relying on statistical tests rather than logical reasoning. But it can be problematic if we use statistics to formalize our beliefs about the world, and lend them a rigor they don't deserve, which can happen when the underlying assumptions being tested are wrong.
Ancestry testing is an example that comes to mind. Our DNA carries signatures of our geographic ancestry. So, you can send a saliva sample to an ancestry testing company, and get back an estimate of the proportion of your ancestors who were African, or European or Native American, say. This is based on the assumption that genetic variants that you carry are from these 'parental populations'. If instead the parental populations used were, say, Vanuatu, Mongolia and Tahiti, you'd be told the proportion of your ancestry that was from each of these ancestral populations.
If done correctly, both sets of results would be statistically robust, but thoroughly dependent on the starting assumptions. In most cases, depending on your ancestry, you'd believe the first and not the second. This is because what we decide is true is often based on our a priori assumptions; this is either, as Hume would have said, a lazy way to identify truth, or it's the correct way, or at least the only way we have. But the important thing is to recognize that our choice of scientific truth is often based on belief, not rigorous scientific methods. Indeed, in the ancestry example, the methods were the same in each case.
We still do use inductive and deductive reasoning in science, of course. We might work under the assumption that, say, all complex traits have a genetic cause, asthma/heart disease/schizophrenia/etc. is a complex disease, asthma/heart disease/schizophrenia/etc. must have a genetic cause. And we design our experiments and our analysis based on this inductive reasoning.
Or, we believe that all observed cases of disease X are caused by a variant in gene Y, John has disease X, so John must have the causal variant in gene Y. And this informs how we carry out our work. And so on. Even though the starting, or concluding, generalizations may be false, this kind of decision-making about causation is done all the time, based on our beliefs about the arguments. We believe the causal link because we want to believe it.
Essentially the statistical test we choose to apply, if passed by the data, is routinely interpreted as telling us that the putative factor (e.g., a genetic variant) causes our outcome....except when it doesn't. We don't put it this way; instead we say it influences the risk of the outcome, and sometimes we can defend this by a known mechanism but all too often statistical association is assumed to mean causation. Does 'affects risk significantly' mean that it affects risk all the time, but only sometimes does the outcome occur? Or do we mean it sometimes affects risk or even causes the outcome, but we don't know which times that is? Or something different from these alternatives? Statistical tests are subjective, correlation is not the same as causation, and we have very solid reasons to know that many traits have so many individually weak causes that will therefore not pass statistical tests and hence remain elusive.
So, in some ways, our understanding of causality isn't measurably better than Al-Ghazali's, when he invoked the invisible hand of God. Ours invokes generalities that are based on our beliefs about cause and effect, with statistical testing lending them a formality that may or may not be valid. The point is not that our working assumptions are false, but that we don't have a way to reliably determine whether or not they are. This also shows why the widely touted criterion of 'falsification' is itself faulty (or false?). We can seem to falsify an hypothesis because of problems in our sample, or data, or the computer program we used to analyze it. Maybe the black swan really isn't a swan!
Indeed, the weaknesses of our approaches suggest that they are false or at least misleading in some fundamental way that we have not yet understood -- or perhaps we're not asking the right sorts of questions, because what we want to know is determined by us, not by Nature. We can never know when there's a black swan we haven't yet seen. Statistical sophistication gives us the illusion that we're past this, and we all know that correlation doesn't imply causation. But sometimes correlation does in fact imply causation, and we still don't really know how to determine when.
Questions about cause and effect in the natural world have been bandied about since long before Aristotle's time. Aristotle, though, believed that the use of the idea of causality must be supported by theory. Without theory, he thought, a thorough and systematic investigation of the world, answering the 'why' questions, was not possible.
His theory, or categories of possible answers to why questions, consisted of these four causes (from the (Stanford Encyclopedia of Philosophy):
The efficient cause may be the only of Aristotle's four types of cause that actually addresses our contemporary idea of cause and effect, explaining what produced an end result. While Aristotle's theorizing about causality is important in the history of science, when we're thinking about cause and effect we aren't generally thinking anymore in terms of what a trait is made of, what it looks like, or what it is 'for', even if these may be descriptive of our trait. 'Because X' may answer why questions, but it doesn't generally lead to understanding of how one event necessarily follows another, or allow us to predict future events, which is a large part of what we're interested in in science. So, Aristotle's efficient cause is probably the one category we still spend time and money on, 'the primary source' of the thing we're interested in.The material cause: “that out of which”, e.g., the bronze of a statue. The formal cause: “the form”, “the account of what-it-is-to-be”, e.g., the shape of a statue. The efficient cause: “the primary source of the change or rest”, e.g., the artisan, the art of bronze-casting the statue, the man who gives advice, the father of the child. The final cause: “the end, that for the sake of which a thing is done”, e.g., health is the end of walking, losing weight, purging, drugs, and surgical tools.
This week's episode of BBC Radio 4 program, In Our Time is a discussion of the life and thought of 12th century Islamic thinker Al-Ghazali. Al-Ghazali wrote about theology, jurisprudence, mysticism and philosophy, interesting in their own right, but what struck us was the discussion of his thinking about causality. His book, The Incoherence of the Philosophers, changed Islamic thinking about epistemology. Al-Ghazali vehemently rejected the material explanations of causation of Aristotle and others, instead proposing that all causality was due to the Will of God.
![]() |
| "Al Ghazzali illustration" by Claud Field - Alchemy of Happiness - Field.djvu. Licensed under Public Domain via Wikimedia Commons |
The example given on In Our Time was of cotton burning when it touches fire. Must the cotton burn? The philosophers who Al-Ghazali rejected would have said yes, but Al-Ghazali believed that the material world must be seen as based on God's relationship to the world. So, causality has to be viewed in relation to God, and from this perspective, the cotton could fail to burn because miracles are always possible. Or, because God makes everything happen, he could stop the burning. Or, yes, fire causes cotton to burn, but only because God intervenes in everything, and makes the cotton burn when it touches fire. It only looks as though there are physical causes to everything. In reality, God is the cause.
Of course, this view is easily dismissible by scientists, who believe in material causes. But science doesn't, in fact, always do much better. As was noted on In Our Time, David Hume, 18th century Scottish philosopher who played a central role in the Enlightenment, the approach to material questions about the world based on empirical, objective observation, had read Al-Ghazali and was influenced in part by what he read.
Hume would have said that, yes, we think that fire must burn cotton, but this is something we believe only out of habit. We've always seen cotton burn when it touches fire, but it's sloppy thinking to assume it will always burn. Indeed, he was a critic of this kind of inductive reasoning, the assumption that nature will always behave as we expect, based on our past observations, and will continue to be regular. The "all the swans we've seen are white, therefore all swans are white" pure induction kind of thinking. The problem, of course, is that the first black swan we see destroys our generality about swans, and brings us back to square one. Or, in practice, we try to cling to our theory and explain the exception (e.g., all white swans are white, the black one is a different species or type).
An alternative to induction is deduction -- "All men are mortal, Socrates is a man, Socrates is mortal." But, deduction relies on a starting generality that is assumed to be correct, and if it is not, the conclusion will be false as well. So, both forms of reasoning have fundamental problems, but both are still in use, in part for lack of better alternatives, and in part because in practice they are often useful. If gravity is universal and based on distance and mass, then we can predict the motion of objects we have never seen before, or we can understand the motion of multiple objects. We use deduction in genetics by reasoning from assumptions about what genes are and do to infer what genetically has caused some outcome (e.g., an instance of a disease). But here we too often don't have strong deductive rigor in our premises.
Since the 19th century and the birth of statistical approaches to evaluating the strength of association between risk factors and outcomes, though, in some ways science has sidestepped the problems with determining causality by relying on statistical tests rather than logical reasoning. But it can be problematic if we use statistics to formalize our beliefs about the world, and lend them a rigor they don't deserve, which can happen when the underlying assumptions being tested are wrong.
Ancestry testing is an example that comes to mind. Our DNA carries signatures of our geographic ancestry. So, you can send a saliva sample to an ancestry testing company, and get back an estimate of the proportion of your ancestors who were African, or European or Native American, say. This is based on the assumption that genetic variants that you carry are from these 'parental populations'. If instead the parental populations used were, say, Vanuatu, Mongolia and Tahiti, you'd be told the proportion of your ancestry that was from each of these ancestral populations.
If done correctly, both sets of results would be statistically robust, but thoroughly dependent on the starting assumptions. In most cases, depending on your ancestry, you'd believe the first and not the second. This is because what we decide is true is often based on our a priori assumptions; this is either, as Hume would have said, a lazy way to identify truth, or it's the correct way, or at least the only way we have. But the important thing is to recognize that our choice of scientific truth is often based on belief, not rigorous scientific methods. Indeed, in the ancestry example, the methods were the same in each case.
We still do use inductive and deductive reasoning in science, of course. We might work under the assumption that, say, all complex traits have a genetic cause, asthma/heart disease/schizophrenia/etc. is a complex disease, asthma/heart disease/schizophrenia/etc. must have a genetic cause. And we design our experiments and our analysis based on this inductive reasoning.
Or, we believe that all observed cases of disease X are caused by a variant in gene Y, John has disease X, so John must have the causal variant in gene Y. And this informs how we carry out our work. And so on. Even though the starting, or concluding, generalizations may be false, this kind of decision-making about causation is done all the time, based on our beliefs about the arguments. We believe the causal link because we want to believe it.
Essentially the statistical test we choose to apply, if passed by the data, is routinely interpreted as telling us that the putative factor (e.g., a genetic variant) causes our outcome....except when it doesn't. We don't put it this way; instead we say it influences the risk of the outcome, and sometimes we can defend this by a known mechanism but all too often statistical association is assumed to mean causation. Does 'affects risk significantly' mean that it affects risk all the time, but only sometimes does the outcome occur? Or do we mean it sometimes affects risk or even causes the outcome, but we don't know which times that is? Or something different from these alternatives? Statistical tests are subjective, correlation is not the same as causation, and we have very solid reasons to know that many traits have so many individually weak causes that will therefore not pass statistical tests and hence remain elusive.
So, in some ways, our understanding of causality isn't measurably better than Al-Ghazali's, when he invoked the invisible hand of God. Ours invokes generalities that are based on our beliefs about cause and effect, with statistical testing lending them a formality that may or may not be valid. The point is not that our working assumptions are false, but that we don't have a way to reliably determine whether or not they are. This also shows why the widely touted criterion of 'falsification' is itself faulty (or false?). We can seem to falsify an hypothesis because of problems in our sample, or data, or the computer program we used to analyze it. Maybe the black swan really isn't a swan!
Indeed, the weaknesses of our approaches suggest that they are false or at least misleading in some fundamental way that we have not yet understood -- or perhaps we're not asking the right sorts of questions, because what we want to know is determined by us, not by Nature. We can never know when there's a black swan we haven't yet seen. Statistical sophistication gives us the illusion that we're past this, and we all know that correlation doesn't imply causation. But sometimes correlation does in fact imply causation, and we still don't really know how to determine when.
Monday, November 3, 2014
TMI? Personalized probability
I went to my dentist on Friday expecting to go home with a permanent crown on one of my molars but I went home with an appointment with an endodontist for evaluation for a root canal instead. My dentist, who I think is a very good one, said that 6% of people getting crowns go on to need a root canal, and, based on my symptoms, he was 90% sure I'd be one of those 6%. Lovely.
But what does that 6% even mean? Just off the top of my head, it could reflect the fallibility of indications for the need for a crown, it could mean that the need for a crown and a root canal are unrelated events that just happened to occur at the same time, it could mean that some dentists are better diagnosticians than others. And of course, it's based on the assumption that all crowns are necessary, and are fixing something real, because if not, people getting a crown who didn't need them aren't actually at risk of a root canal. Maybe all of these. Someone might know, but not I. That 90% was, I assume, mostly professional courtesy, allowing for the possibility that the endodontist may not agree--that is, it was a subjective but not really a probability statement and how he arrived at it is anyone's guess (maybe even his own!).
My dentist also told me that there is probably a subgroup of crown patients that is more likely to need root canals than others. If I remember correctly what he said -- keep in mind that I had just been told I'd need one of the most dreaded dental procedures there is, so my memory could be faulty here -- he said that the subgroup consists of people who experienced more pain before the crown procedure than the other 94%. Which I translated to mean that these could be people with problems that their dentists misdiagnosed; maybe they needed a root canal from the beginning, but I could be totally wrong about that because I really know next to nothing about how dentists decide when a tooth needs a crown or a root canal. In any case, it did make me think about the whole idea of probability and prediction.
First, I wasn't at all surprised that I landed in that 6% group. My family has experienced what we think is an unlikely number of rare events, so of course I would be in that 6%. Right? Well, no. Each event has been seemingly independent, with separate causes, so what has happened before has no bearing on whether or not I'm going to have a root canal. One person being hit by lightning (say) doesn't make it more likely that another will be hit by a bus. Unless, you know, bad luck is genetic.
But let's say these things aren't genetic. Let's say they are random events and everyone at risk is equally at risk. It's easy to calculate the chances of something like winning the lottery because the denominator, the number of people at risk, is readily available -- it's the number of people who buy a lottery ticket. It's a lot harder to calculate the chances of being hit by lightning or by a bus because we can't know how many people are at risk. We'd have to know how many people are outside during a storm, or crossing a street when a bus comes along, and that's impossible unless we're asking about your street corner or my backyard.
In that case, we've made it easy to calculate the denominator, but the likelihood of the event happening exactly there has now become astronomically low, no matter how precisely we know how many are at risk. And if we expand the geographic area again to a city or a state so that the event has a reasonable probability of happening there, we might know how many people are struck by lightning or a bus (though, that's a hard one too because surely people don't always report being hit by a bus) but again, we've lost sight of the denominator so it's impossible to calculate the probabilities. Unless, like the National Weather Service, we assume the whole population of the US is at risk every year, in which case, according to their calculations, each of us has a probability of being hit by lightning in a given year ranging from 1/1,900,000 to 1/960,000 depending on whether we're using actual reported lightning strikes or estimated strikes.
What are the odds of being hit by a bus in a given year? This is a hard statistic to find, and it's necessarily just an estimate not only because such accidents aren't always reported, and we haven't specified whether we mean fatal or not, but also because sometimes bus accidents involve people inside the bus. Given all that, the odds of being hit by a bus per year in the UK have been estimated to be 1/500,000. That would mean something like that there were x many reported bus accidents and so many people in the UK in that year (but are they counting tourists and undocumented workers?). But that's also a bit suspect, because some people live and walk around in cities, and others in the country where there are no buses. Or what if they drive? Does being hit by a bus while in one's car count in the reported statistic?
Let's say someone in your family has been hit by a bus, and another family member has been hit by lightning. If the two are independent -- and how would we know that? -- then the odds would be the product of the two figures given above, roughly ten to the minus 11th power, or 11 zeroes after the decimal point. Since there are only about 10 to the 9th power humans on earth, your nuclear family is probably the only family on Earth in which these two events have occurred together (actually, this is impossible to estimate for a given family if we don't specify family size and so on, but the point is that there can be events so rare that the fact of both of them happening in one family is a one-time occurrence.) If we excuse indigenous people living where there are no buses (there is likely to be lightning), even just these two events make you super-unique. And it's such a rare circumstance that it's hard to know what it has to teach us about probabilities of such things happening. Most of the useful stuff we learn from science is based on reproducible events -- because if not, the odd thing that happened ends up on the News of the Weird page in your local paper. But even things like lightning strikes are not truly reproducible.
We can enumerate all the steps that led to my Uncle Ed being hit by a bus, but they won't necessarily apply to your Uncle Ed's experience with a bus. The outcome, that they both ended up in front of a bus at the wrong time, is the only generalizable and useful fact here. That my Uncle Ed stopped to buy a newspaper to read this week's News of the Weird, and was so absorbed in the story about the elephant that he stepped off the curb to cross the street just when the bus came around the corner is irrelevant to the facts of your Uncle Ed's ill-fated meeting with his bus. Your Uncle Ed, as it happened, was in a hurry and, yes, texting, and he stepped off the curb to get around the throng of tourists gawking at the Empire State Building just as a bus came along.
The facts in each case, interesting and important as they are to each Uncle Ed and his family, can't be used to predict the steps by which someone else's Uncle Ed will end up being hit by a bus. We might conclude that reading the News of the Weird is dangerous, but that seems illogical. We might, but probably shouldn't conclude that tourists are dangerous. But if we begin to collect enough stories about people being hit by a bus while texting, we can start to suspect that texting might be dangerous. But by looking at causation in aggregate, we're likely to pay attention to common causes and ignore the rare ones. Indeed, my dentist didn't feel he needed to tell me that my tooth that seemed to need a simple crown might end up also needing a root canal.
But, in every case, we've got a lot of information that doesn't apply to everyone else, or even anyone else, even if it applies in the individual case. TMI, except that if the event is rare -- and particularly if it's unique -- we can't always determine which information is relevant to its cause, especially if there are multiple pathways to the outcome. It's only when we've got a lot of replications of the same event that we can begin to sort that out, but even then we usually can't begin to predict the event. Not everyone texting gets hit by a bus, just as not everyone who smokes gets cancer, and not everyone with a BRCA risk allele gets breast cancer. Public health policy warns against the effects of smoking anyway, even when only 10% of smokers get cancer. We're not good at predicting who will be among that 10%, so we warn everyone.
The danger of numerology
We often read about coincidences like the Uncle Eds'. Probably, you can find the story in the News of the Weird -- two Uncle Eds hit by a bus. And if it was the same bus, all the better. But there is a very common danger in that. If you sort through enough possible events, what in itself seems like an incredible coincidence becomes essentially expectable. There are enough rare events in this world that the chance two Eds share at least some such event is not so small. The chances that my family has experienced the particular rare events that we've experienced is probably so small that we're the only family ever to have experienced these specific events in the history of families on Earth, but the number of families that are unique in the same way is in fact probably quite common. And depending on how we classify events, every family is unique. (A good discussion of this is in a recent podcast of a BBC Radio4 program More or Less, that tries to debunk misuses of statistics.) We see the problem of assessing the uniqueness of events all the time in twin studies, too. If my dog and your dog have the same name it's just a curious coincidence, but if my dog and my twin's dog have the same name, Nature will publish that as a genetic finding (thank you, Holly). So we have to be careful of that sort of numerological game.
The nature of prediction, done properly
The concept of 'probability' is a very elusive one, and really it has no single definition. But most uses refer to repeatable events, as mentioned above. That is, the probability of an outcome is the fraction of times that outcome is observed among repeated experiences. Likewise, prediction is generally based on aggregate data. That's what epidemiology does; collect a pile of data in a population and figure out which variables, or exposures, are most likely associated with an outcome or disease. In genetic epidemiology, if enough people with a disease share a genetic variant, that variant starts to look causal. In the easiest cases, which are usually rare diseases, the variant can then often, but certainly not always, be considered predictive, or at least diagnostic.
As we've said many times before, epidemiology and many other areas of observational science do best when causation is simple. Simple point, or single-factor causes, make the notion of replication more sensible and more reliably observable. But this is far less clear when causation is complex, involving multiple simultaneous contributing factors, no one combination of which is necessary.
Too often, even if hundreds of millions of dollars have been spent on trying to understand the cause of a disease, prediction is hard. In a very nice post on the blog 'chronic health, Luzt Kraushaar shows that this is true for heart disease. He writes:
But we're all rare events. Every genome that has ever existed has been unique. We are each now a 1 in 7.5 billion event (and that's without counting our also entirely unique ancestors). And this means that every event for everyone of us follows its own path. Even the phenotype of single gene diseases can differ between individuals because while they may share a causal variant it's in a unique genomic context, and the modifying effects of that context are unique. Further, when a disease involves environmental exposures, everyone's will be unique, and this makes retrodicting cause (which is what epidemiology does) difficult, but predicting it is even more difficult because we can't predict future environments, and we can't know which particular genomic contexts, each one of them unique, are significant risk factors or what kinds of modifiers they'll be.
So, while we're being promised miracles by believers in personalized medicine based on our DNA, this in some ways will be too much information that we don't know what to make of. Predicting complex diseases, those due to environmental exposures interacting with many genes, is going to be impossible, in principle, much of the time. Who knew News of the Weird was going to cause Uncle Ed to walk in front of a bus?
![]() |
| Root canal; Wikipedia |
My dentist also told me that there is probably a subgroup of crown patients that is more likely to need root canals than others. If I remember correctly what he said -- keep in mind that I had just been told I'd need one of the most dreaded dental procedures there is, so my memory could be faulty here -- he said that the subgroup consists of people who experienced more pain before the crown procedure than the other 94%. Which I translated to mean that these could be people with problems that their dentists misdiagnosed; maybe they needed a root canal from the beginning, but I could be totally wrong about that because I really know next to nothing about how dentists decide when a tooth needs a crown or a root canal. In any case, it did make me think about the whole idea of probability and prediction.
First, I wasn't at all surprised that I landed in that 6% group. My family has experienced what we think is an unlikely number of rare events, so of course I would be in that 6%. Right? Well, no. Each event has been seemingly independent, with separate causes, so what has happened before has no bearing on whether or not I'm going to have a root canal. One person being hit by lightning (say) doesn't make it more likely that another will be hit by a bus. Unless, you know, bad luck is genetic.
But let's say these things aren't genetic. Let's say they are random events and everyone at risk is equally at risk. It's easy to calculate the chances of something like winning the lottery because the denominator, the number of people at risk, is readily available -- it's the number of people who buy a lottery ticket. It's a lot harder to calculate the chances of being hit by lightning or by a bus because we can't know how many people are at risk. We'd have to know how many people are outside during a storm, or crossing a street when a bus comes along, and that's impossible unless we're asking about your street corner or my backyard.
In that case, we've made it easy to calculate the denominator, but the likelihood of the event happening exactly there has now become astronomically low, no matter how precisely we know how many are at risk. And if we expand the geographic area again to a city or a state so that the event has a reasonable probability of happening there, we might know how many people are struck by lightning or a bus (though, that's a hard one too because surely people don't always report being hit by a bus) but again, we've lost sight of the denominator so it's impossible to calculate the probabilities. Unless, like the National Weather Service, we assume the whole population of the US is at risk every year, in which case, according to their calculations, each of us has a probability of being hit by lightning in a given year ranging from 1/1,900,000 to 1/960,000 depending on whether we're using actual reported lightning strikes or estimated strikes.
What are the odds of being hit by a bus in a given year? This is a hard statistic to find, and it's necessarily just an estimate not only because such accidents aren't always reported, and we haven't specified whether we mean fatal or not, but also because sometimes bus accidents involve people inside the bus. Given all that, the odds of being hit by a bus per year in the UK have been estimated to be 1/500,000. That would mean something like that there were x many reported bus accidents and so many people in the UK in that year (but are they counting tourists and undocumented workers?). But that's also a bit suspect, because some people live and walk around in cities, and others in the country where there are no buses. Or what if they drive? Does being hit by a bus while in one's car count in the reported statistic?
Let's say someone in your family has been hit by a bus, and another family member has been hit by lightning. If the two are independent -- and how would we know that? -- then the odds would be the product of the two figures given above, roughly ten to the minus 11th power, or 11 zeroes after the decimal point. Since there are only about 10 to the 9th power humans on earth, your nuclear family is probably the only family on Earth in which these two events have occurred together (actually, this is impossible to estimate for a given family if we don't specify family size and so on, but the point is that there can be events so rare that the fact of both of them happening in one family is a one-time occurrence.) If we excuse indigenous people living where there are no buses (there is likely to be lightning), even just these two events make you super-unique. And it's such a rare circumstance that it's hard to know what it has to teach us about probabilities of such things happening. Most of the useful stuff we learn from science is based on reproducible events -- because if not, the odd thing that happened ends up on the News of the Weird page in your local paper. But even things like lightning strikes are not truly reproducible.
We can enumerate all the steps that led to my Uncle Ed being hit by a bus, but they won't necessarily apply to your Uncle Ed's experience with a bus. The outcome, that they both ended up in front of a bus at the wrong time, is the only generalizable and useful fact here. That my Uncle Ed stopped to buy a newspaper to read this week's News of the Weird, and was so absorbed in the story about the elephant that he stepped off the curb to cross the street just when the bus came around the corner is irrelevant to the facts of your Uncle Ed's ill-fated meeting with his bus. Your Uncle Ed, as it happened, was in a hurry and, yes, texting, and he stepped off the curb to get around the throng of tourists gawking at the Empire State Building just as a bus came along.
![]() |
| New York City bus; Wikipedia |
The facts in each case, interesting and important as they are to each Uncle Ed and his family, can't be used to predict the steps by which someone else's Uncle Ed will end up being hit by a bus. We might conclude that reading the News of the Weird is dangerous, but that seems illogical. We might, but probably shouldn't conclude that tourists are dangerous. But if we begin to collect enough stories about people being hit by a bus while texting, we can start to suspect that texting might be dangerous. But by looking at causation in aggregate, we're likely to pay attention to common causes and ignore the rare ones. Indeed, my dentist didn't feel he needed to tell me that my tooth that seemed to need a simple crown might end up also needing a root canal.
But, in every case, we've got a lot of information that doesn't apply to everyone else, or even anyone else, even if it applies in the individual case. TMI, except that if the event is rare -- and particularly if it's unique -- we can't always determine which information is relevant to its cause, especially if there are multiple pathways to the outcome. It's only when we've got a lot of replications of the same event that we can begin to sort that out, but even then we usually can't begin to predict the event. Not everyone texting gets hit by a bus, just as not everyone who smokes gets cancer, and not everyone with a BRCA risk allele gets breast cancer. Public health policy warns against the effects of smoking anyway, even when only 10% of smokers get cancer. We're not good at predicting who will be among that 10%, so we warn everyone.
The danger of numerology
We often read about coincidences like the Uncle Eds'. Probably, you can find the story in the News of the Weird -- two Uncle Eds hit by a bus. And if it was the same bus, all the better. But there is a very common danger in that. If you sort through enough possible events, what in itself seems like an incredible coincidence becomes essentially expectable. There are enough rare events in this world that the chance two Eds share at least some such event is not so small. The chances that my family has experienced the particular rare events that we've experienced is probably so small that we're the only family ever to have experienced these specific events in the history of families on Earth, but the number of families that are unique in the same way is in fact probably quite common. And depending on how we classify events, every family is unique. (A good discussion of this is in a recent podcast of a BBC Radio4 program More or Less, that tries to debunk misuses of statistics.) We see the problem of assessing the uniqueness of events all the time in twin studies, too. If my dog and your dog have the same name it's just a curious coincidence, but if my dog and my twin's dog have the same name, Nature will publish that as a genetic finding (thank you, Holly). So we have to be careful of that sort of numerological game.
The nature of prediction, done properly
The concept of 'probability' is a very elusive one, and really it has no single definition. But most uses refer to repeatable events, as mentioned above. That is, the probability of an outcome is the fraction of times that outcome is observed among repeated experiences. Likewise, prediction is generally based on aggregate data. That's what epidemiology does; collect a pile of data in a population and figure out which variables, or exposures, are most likely associated with an outcome or disease. In genetic epidemiology, if enough people with a disease share a genetic variant, that variant starts to look causal. In the easiest cases, which are usually rare diseases, the variant can then often, but certainly not always, be considered predictive, or at least diagnostic.
As we've said many times before, epidemiology and many other areas of observational science do best when causation is simple. Simple point, or single-factor causes, make the notion of replication more sensible and more reliably observable. But this is far less clear when causation is complex, involving multiple simultaneous contributing factors, no one combination of which is necessary.
Too often, even if hundreds of millions of dollars have been spent on trying to understand the cause of a disease, prediction is hard. In a very nice post on the blog 'chronic health, Luzt Kraushaar shows that this is true for heart disease. He writes:
If weather forecasts were as reliable as cardiovascular risk prediction tools, meteorologists would miss two thirds of all hurricanes, expect rain for 8 out of 10 sunny days, and fail to see the parallels to fortune telling.Epidemiologists have tried for decades to separate the wheat from the chaff with heart disease, to determine which risk factors are predictive; which are necessary and sufficient. In Uncle Ed terms, this means they've tried to determine whether reading News of the Weird is causal or whether the cause is actually texting. Or maybe both are causal. And even though heart disease is a major cause of death in rich countries, we still don't know very well how to predict who'll be affected. In a way, that's because every heart attack is a rare, even unique event, and aggregating data may discard the important variables about the cause of my heart attack. For my Uncle Ed, the News of the Weird was an important cause of his getting hit by the bus, but not predictive for anyone else, or at least not reliably predictive.
But we're all rare events. Every genome that has ever existed has been unique. We are each now a 1 in 7.5 billion event (and that's without counting our also entirely unique ancestors). And this means that every event for everyone of us follows its own path. Even the phenotype of single gene diseases can differ between individuals because while they may share a causal variant it's in a unique genomic context, and the modifying effects of that context are unique. Further, when a disease involves environmental exposures, everyone's will be unique, and this makes retrodicting cause (which is what epidemiology does) difficult, but predicting it is even more difficult because we can't predict future environments, and we can't know which particular genomic contexts, each one of them unique, are significant risk factors or what kinds of modifiers they'll be.
So, while we're being promised miracles by believers in personalized medicine based on our DNA, this in some ways will be too much information that we don't know what to make of. Predicting complex diseases, those due to environmental exposures interacting with many genes, is going to be impossible, in principle, much of the time. Who knew News of the Weird was going to cause Uncle Ed to walk in front of a bus?
Tuesday, October 21, 2014
And it's even worse for Big Data.....
By
Ken Weiss
Last week we pointed out that the history of technology enabled larger and more extensive and exhaustive enumeration of genomic variation, to apply to understanding the cause of human traits, important diseases or even normal variation like the recent stature paper.
We basically noted that even the earlier, cruder methods such as 'linkage' and 'candidate gene' analysis did not differ logically in their approach as much as advocates of mega-studies often allege, but more importantly, all the approaches have for decades told the same story. That is, that from a genomic causal point of view, with a few exceptions, complex traits that don't seem to be due to strongly-causal simple genotypes aren't. Yet the profession has been vigorously insisting on ever bigger and more costly, long-term, too-big-to stop studies to look for some vaguely specified pot of tractably simple causation at the end of the genomic rainbow.
But the problem is even worse. The history we reviewed in last week's post did not include other factors that make the problem even more complex. First, is somatic mutation, about which we wrote separately last week. But environment and lifestyle exposures are far more important to the causation of most traits than genomic variation. The obvious and rather clear evidence is that the heritability is almost always quite modest. With a few exceptions like stature (which is only an exception if you standardize height for cohort or age, that is, for differing environments), most traits of interest have heritability on the order of 30% or so.
But why are environmental contributions problematic? If we can Next-Gen our way to enumerating all the (germline) genomic variants, why can't someone invent a Next-Env assessor of environments? There are several reasons, each profound and challenging to say the least.
More on history
In last Friday's (Oct 17) post on Big Data, we noted that history had shown that even cruder messages were successful at giving us the basic message we sought, even if it wasn't the message we wanted to find.
The history of major, long-term, large-scale studies in general shows another reason we invest false expectations in the desire for Big Data. Forgetting physics and other sciences, there have been a number of very big, long-term studies of various types. We have had registries, studies of radiation exposure, follow-ups on diets, drug, hormone and other usage, studies on lipids in places like Framlingham, sometimes multi-generational follow-ups. They have made some findings, sometimes important findings, but usually two things have happened. First, they quickly reached the point of diminishing returns, even if the investigators pressed to keep the studies going. Second, many if not perhaps even most of the major findings have turned out, years or decades later, even in the same study, not to hold up very well.
This isn't the investigators' fault. It's the fault of the idea itself. Even things like, say, lipids and other such risk factors are estimated in different ways later in a study than they were earlier, as ideas about what to measure evolves. Measurement technology evolves so that earlier data are not considered reliable or accurate or cogent enough. The repeated desire is to re-survey or re-measure, but then all sorts of other things happen to the study population (not least of which is that subjects die and can't be recontacted). Yet, and we've all heard this excuse for renewal in Study Section meetings, "after all that we've invested, it would be a shame to stop now!"
In a very real scientific investment sense, the shame in many such cases is not to stop now. That involves some very difficult politics of various kinds, and of course scientists are people, too, and we don't just react to the facts as we clearly know them.
Is an Awakening in progress?
Here and there (not just here on MT!) are signs that not only is the complexity being openly recognized as a problem, but perhaps there's an awakening or a formal recognition of the important lessons of Big Data.
Although the advocates of Big Data Evermore surely remain predominant, a number of Tweets out of this week's American Society of Human Genetics meetings suggest at least a recognition among some that caution is warranted when it comes to the promises of Big Data. The dawning has not yet, we think, reached the serious recognition that Big Data is a wastefully costly and misleading way to go, but at least that genotype-based predictions have been over-promised. How much more time and intellectual as well as fiscal resources will be dispersed to the winds before a serious re-thinking of the problem or the goals occurs? Or is the technology first approach too entrenched in current ways of scientific thinking for anything like such a change of direction to happen?
We basically noted that even the earlier, cruder methods such as 'linkage' and 'candidate gene' analysis did not differ logically in their approach as much as advocates of mega-studies often allege, but more importantly, all the approaches have for decades told the same story. That is, that from a genomic causal point of view, with a few exceptions, complex traits that don't seem to be due to strongly-causal simple genotypes aren't. Yet the profession has been vigorously insisting on ever bigger and more costly, long-term, too-big-to stop studies to look for some vaguely specified pot of tractably simple causation at the end of the genomic rainbow.
![]() |
| When a genetic effect is strong, we can usually find it, in family members; Wikipedia |
But the problem is even worse. The history we reviewed in last week's post did not include other factors that make the problem even more complex. First, is somatic mutation, about which we wrote separately last week. But environment and lifestyle exposures are far more important to the causation of most traits than genomic variation. The obvious and rather clear evidence is that the heritability is almost always quite modest. With a few exceptions like stature (which is only an exception if you standardize height for cohort or age, that is, for differing environments), most traits of interest have heritability on the order of 30% or so.
But why are environmental contributions problematic? If we can Next-Gen our way to enumerating all the (germline) genomic variants, why can't someone invent a Next-Env assessor of environments? There are several reasons, each profound and challenging to say the least.
1. Identifying all the potentially major environmental contributors, which may occur only sporadically, decades before someone is surveyed, unbeknownst to the person, or simply unknown to the investigator, is at least a largely insuperable challenge.
2. Defining the exposure is difficult and measurement error probably large and to an unknown or even unknowable extent. We might refine our list, but how accurately can we really expect to measure subtle things like behavior habits in early childhood, exposures in utero, or even such seemingly obvious things like fat intake, or even smoking exposure.
3. Somatic mutations will interact with environmental factors just as germline ones will, so our models of causation are simply inaccurate to an unknown extent.
4. Suppose we could get perfect recall, and perfect measurement, of all exposure variables that affected our sampled individuals from in utero to their current ages (and, assuming pre-conception effects on gene expression by such things as epigenetic marking). That would enable accurate retrospective data fitting, but there is simply no way to predict a current person's future exposures. The agents and exposure levels, and even identity of risk factors is simply unknowable, even in principle.These facts, and they are relevant facts, show that no matter the DNA sequencing sophistication, we cannot expect to turn complex traits into simple traits, or even enumerable, predictable ones.
More on history
In last Friday's (Oct 17) post on Big Data, we noted that history had shown that even cruder messages were successful at giving us the basic message we sought, even if it wasn't the message we wanted to find.
The history of major, long-term, large-scale studies in general shows another reason we invest false expectations in the desire for Big Data. Forgetting physics and other sciences, there have been a number of very big, long-term studies of various types. We have had registries, studies of radiation exposure, follow-ups on diets, drug, hormone and other usage, studies on lipids in places like Framlingham, sometimes multi-generational follow-ups. They have made some findings, sometimes important findings, but usually two things have happened. First, they quickly reached the point of diminishing returns, even if the investigators pressed to keep the studies going. Second, many if not perhaps even most of the major findings have turned out, years or decades later, even in the same study, not to hold up very well.
This isn't the investigators' fault. It's the fault of the idea itself. Even things like, say, lipids and other such risk factors are estimated in different ways later in a study than they were earlier, as ideas about what to measure evolves. Measurement technology evolves so that earlier data are not considered reliable or accurate or cogent enough. The repeated desire is to re-survey or re-measure, but then all sorts of other things happen to the study population (not least of which is that subjects die and can't be recontacted). Yet, and we've all heard this excuse for renewal in Study Section meetings, "after all that we've invested, it would be a shame to stop now!"
In a very real scientific investment sense, the shame in many such cases is not to stop now. That involves some very difficult politics of various kinds, and of course scientists are people, too, and we don't just react to the facts as we clearly know them.
Is an Awakening in progress?
Here and there (not just here on MT!) are signs that not only is the complexity being openly recognized as a problem, but perhaps there's an awakening or a formal recognition of the important lessons of Big Data.
Although the advocates of Big Data Evermore surely remain predominant, a number of Tweets out of this week's American Society of Human Genetics meetings suggest at least a recognition among some that caution is warranted when it comes to the promises of Big Data. The dawning has not yet, we think, reached the serious recognition that Big Data is a wastefully costly and misleading way to go, but at least that genotype-based predictions have been over-promised. How much more time and intellectual as well as fiscal resources will be dispersed to the winds before a serious re-thinking of the problem or the goals occurs? Or is the technology first approach too entrenched in current ways of scientific thinking for anything like such a change of direction to happen?
Monday, September 29, 2014
Predicting future environments: it's impossible in principle
We often caution here on MT about predicting the future. There's an idea afoot that once we know our genomes we'll be able to predict who will get what chronic disease, even when the disease is the result of genes interacting with environmental factors. Or even multiple genes and multiple environments. But, as we've often said, making such predictions with known accuracy is impossible in principle. And here's why.
A BBC television series in the 1970's, "A Taste of Britain," hosted by Derek Cooper, set out to document traditional foods and food practices in Britain before they disappeared. The long-running BBC Radio 4 show, "The Food Programme," revisits this series by traveling to the places documented 40 years ago to find out what became of the foods, the people, and the traditions the original program had captured on film. "A Taste of Britain Revisited" is a fascinating glimpse into traditional foods and how they've fared in the intervening years.
Cooper urgently tracked down cockle pickers, sewin fishermen, truffle hunters, cheese makers, colliers wives, makers of Yorkshire Pudding, makers of Welsh cakes and laverbread, and more, certain that the traditions he was documenting would soon be lost.
And some were. Once miners were moved into council housing in Yorkshire, away from coal-fired ovens, Yorkshire Pudding was never the same. Native English crayfish will soon be extinct because a heartier North American variety has taken over, bringing a virus that kills the natives with it. Cooper filmed the last traditional hearth in Wales, which was replaced but something more modern just weeks later.
And some have changed. The traditional Dorset Blue Vinny cheese that Cooper was told he was eating was in fact a second-rate Stilton, the real thing not having been made since World War II when the making of soft and semi-soft cheeses was banned, and only hard, transportable cheeses like cheddar could be made, to be sent to the cities. But the recipe for Blue Vinny was found in the 1980's, and the cheese revived. Cockles were in short supply in the 1970's and had to be imported. The numbers increased for a while after that, fell and increased again, but now aren't as plentiful as they were, and cockle gatherers don't go to the beach by horse and cart any longer.
And some have remained the same, but the economics have changed. Welsh laver, or seaweed, is now being sold in Japan, when just 40 years ago getting it to London was a major accomplishment. Businesses have been bought up by foreign companies. A Spanish company now buys British cockles, and they are now being canned and sold in Spain. "Cans is something we never thought we'd put cockles into, but if that's the way they want them, that's the way they can have them."
So, Cooper wasn't entirely right that he was documenting the final days of many traditions and foods, though he was certainly documenting change. I don't know what he thought was going to replace traditional foods, but perhaps the biggest change over the last 40 years was the rise of processed foods. Whether or not he foresaw that I don't know, but even if he or anyone else had, the health effects couldn't have been predicted. Indeed, even with all the evidence before us, we can't really say what the health effects of any of the specific changes have been. Yes, people are more obese, have more heart disease, stroke, hypertension, cancers and so on, but in a very real sense, a major 'cause' of these diseases is control of infectious diseases.
But, that's not the real point of this post. The point is that it looks as though the common, late onset chronic diseases we are dying of now are the result of complex interactions between genes and environment. Epidemiologists have been trying to identify environmental risk factors for decades, with only modest success -- meaning that it's not clear that we know which aspects of our environments are linked with risk. But even if we did, these two series of programs on traditional foods in Britain make it clear that while we may fit today's disease cases to yesterday's exposures, it's impossible to predict what people will be eating not far into the future, so that even if we do identify risk alleles and risky environments, we don't know that carriers of that allele will still be exposed to risky environments not many years from now. So, we can't predict their disease with more than retrospective accuracy, when we know from experience that risks of the same traits change rapidly and substantially.
Here's an example. Type 2 diabetes has risen to epidemic levels in Native American populations, and Mexican Americans who are admixed with Native Americans, since World War II. The nature of the disease and the pattern of the epidemic suggests that a fairly simple genetic background may be responsible. But, even if so, that's only in the provocative environment of the last 60 or so years. That's because there was little to no type 2 diabetes in Native Americans, or Mexican Americans, before then, yet whichever alleles are responding now, leading to glucose intolerance and so forth have basically not changed in frequency. It's the environment -- diet and exercise, presumably -- that has changed. This epidemic could simply not have been predicted 60 years ago, even if we'd been genotyping people then, because whatever risk alleles are now responsible were not causing disease before the environment changed.
So, the promise that genotyping every infant at birth will allow us to predict the late onset, complex diseases they will eventually have is unlikely to be met. Instead, for many or most complex traits, it's an illusion that's being sold as fact.
A BBC television series in the 1970's, "A Taste of Britain," hosted by Derek Cooper, set out to document traditional foods and food practices in Britain before they disappeared. The long-running BBC Radio 4 show, "The Food Programme," revisits this series by traveling to the places documented 40 years ago to find out what became of the foods, the people, and the traditions the original program had captured on film. "A Taste of Britain Revisited" is a fascinating glimpse into traditional foods and how they've fared in the intervening years.
Cooper urgently tracked down cockle pickers, sewin fishermen, truffle hunters, cheese makers, colliers wives, makers of Yorkshire Pudding, makers of Welsh cakes and laverbread, and more, certain that the traditions he was documenting would soon be lost.
![]() |
| Cockle pickers, Wales: Daily Mail, 2008 |
![]() |
| Fishing for sewin (sea trout) in traditional coracles, Wales; Wikipedia |
And some have changed. The traditional Dorset Blue Vinny cheese that Cooper was told he was eating was in fact a second-rate Stilton, the real thing not having been made since World War II when the making of soft and semi-soft cheeses was banned, and only hard, transportable cheeses like cheddar could be made, to be sent to the cities. But the recipe for Blue Vinny was found in the 1980's, and the cheese revived. Cockles were in short supply in the 1970's and had to be imported. The numbers increased for a while after that, fell and increased again, but now aren't as plentiful as they were, and cockle gatherers don't go to the beach by horse and cart any longer.
![]() |
| Horse and Cart on Beach, William Ayerst Ingram; Wikipedia |
And some have remained the same, but the economics have changed. Welsh laver, or seaweed, is now being sold in Japan, when just 40 years ago getting it to London was a major accomplishment. Businesses have been bought up by foreign companies. A Spanish company now buys British cockles, and they are now being canned and sold in Spain. "Cans is something we never thought we'd put cockles into, but if that's the way they want them, that's the way they can have them."
So, Cooper wasn't entirely right that he was documenting the final days of many traditions and foods, though he was certainly documenting change. I don't know what he thought was going to replace traditional foods, but perhaps the biggest change over the last 40 years was the rise of processed foods. Whether or not he foresaw that I don't know, but even if he or anyone else had, the health effects couldn't have been predicted. Indeed, even with all the evidence before us, we can't really say what the health effects of any of the specific changes have been. Yes, people are more obese, have more heart disease, stroke, hypertension, cancers and so on, but in a very real sense, a major 'cause' of these diseases is control of infectious diseases.
But, that's not the real point of this post. The point is that it looks as though the common, late onset chronic diseases we are dying of now are the result of complex interactions between genes and environment. Epidemiologists have been trying to identify environmental risk factors for decades, with only modest success -- meaning that it's not clear that we know which aspects of our environments are linked with risk. But even if we did, these two series of programs on traditional foods in Britain make it clear that while we may fit today's disease cases to yesterday's exposures, it's impossible to predict what people will be eating not far into the future, so that even if we do identify risk alleles and risky environments, we don't know that carriers of that allele will still be exposed to risky environments not many years from now. So, we can't predict their disease with more than retrospective accuracy, when we know from experience that risks of the same traits change rapidly and substantially.
Here's an example. Type 2 diabetes has risen to epidemic levels in Native American populations, and Mexican Americans who are admixed with Native Americans, since World War II. The nature of the disease and the pattern of the epidemic suggests that a fairly simple genetic background may be responsible. But, even if so, that's only in the provocative environment of the last 60 or so years. That's because there was little to no type 2 diabetes in Native Americans, or Mexican Americans, before then, yet whichever alleles are responding now, leading to glucose intolerance and so forth have basically not changed in frequency. It's the environment -- diet and exercise, presumably -- that has changed. This epidemic could simply not have been predicted 60 years ago, even if we'd been genotyping people then, because whatever risk alleles are now responsible were not causing disease before the environment changed.
So, the promise that genotyping every infant at birth will allow us to predict the late onset, complex diseases they will eventually have is unlikely to be met. Instead, for many or most complex traits, it's an illusion that's being sold as fact.
Monday, January 13, 2014
The rise and fall and rise again of Mendelism
One hundred and fifty years ago, Moravian monk Gregor Mendel brought the world Mendelian inheritance. That is, the Law of Segregation and the Law of Independent Assortment, among other related principles. We took from his work the idea that we (and our very distant pea-plant relatives) each carry two copies of every gene and each parent randomly passes a single copy to his or her offspring. And, genes are transmitted roughly independently of each other in the formation of sperm and eggs (this isn't strictly true for genes that are close to each other on a chromosome, and the sex chromosomes are a bit different--but these exceptions are irrelevant for this post).
These principles were derived from Mendel's work with pea plants, a study of carefully chosen traits that behaved in a particular way that he felt would be useful for forming better agricultural hybrid plants -- when he crossed plants with different flower color or seed characteristics etc., the traits did not 'blend' in subsequent generations but instead remained discrete: the offspring had one flower color and one type of seed coat; green peas crossed with yellow ones didn't produce in-between colored peas. But, Mendel knew very well that not all traits behaved in this way; he just felt they weren't of interest for his purposes. And without actually saying so, he looked for simple patterns of inheritance that fit other 'hot' science of his time, the development of the atomic theory of matter.
In 1854, in preparation for his later work, Mendel acquired 34 varieties of pea seed, Pisum sativum, from local nurseries. He spent the next 2 years growing them to determine which characters bred true; that is, which parental traits would appear, unblended, in the offspring. Not all did. He chose 7 that did to follow through multiple generations. He produced 7 crosses, each repeated twice, from a seed plant with a particular trait and the second time with that same trait from the pollen plant. In this way he determined that the male and female contribute equally to the offspring.
He bred his plants for the next 8 years, counting the occurrence of the 7 traits in nearly 30,000 peas. Because he was astutely brilliant and chose his traits well, he was able to document what became known as recessive and dominant inheritance, as well as the random segregation of alleles in offspring, and the independent inheritance of genes for traits using a sample size large enough to produce convincing results. In essence he developed statistical rules for the apparent inheritance of his chosen traits. But, he knew he wasn't explaining all of inheritance. And later work on plants that didn't behave according to his laws further confirmed that he hadn't explained everything (sometimes to his frustration, or even despair, though he had moved on to monastery administration by then).
The story of how Mendel's work was largely ignored for 40 years, and rediscovered in 1900 is well-known (we blogged about it some time ago, too, here and here, e.g.). As genetics got under way in earnest in the next few decades, to a sufficient extent, given the technology of the time, Mendelian rules, indeed, seemed rather like 'laws' of nature, although it was clear they didn't explain everything.
One of the most influential and productive geneticists of the early era was Thomas Morgan. He and his students at Columbia created the first linkage maps in fruit flies, identifying chromosomal loci associated with traits in the 1910's, and so on. Even as they did so, Morgan was aware that most traits were due to many genes, and that it was likely that most genes do more than one thing. He wrote in 1917, e.g.:
Morgan also knew that the genetic effect -- the 'trait'-- that he was studying could be restricted to just some developmental stage of his flies, but his point was not the trait but tracing its behavior relative to the idea that they had causal effects that were localized to particular places on chromosomes. The trait-complexity was something he explicitly said needed to wait for a later time when more was known.
A student of Morgan's, A.H. Sturtevant, in his A History of Genetics (1965), wrote:
We bring all this up because an MT reader, André Comeau, plant geneticist and pathologist, recently wrote to Ken with some thoughts on Canadian plant breeder (and flutist) Charles Saunders, developer of Marquis wheat, the most important variety of wheat of the last century in Canada. Saunders developed this wheat after a lot of cross-breeding of different wheats from around the world. Marquis wheat became commercially available in 1909 and because it had a shorter growing season than other available wheat varieties, by 1920, 90% of the wheat grown in Canada was Marquis.
In 1902, Saunders attended the lecture at which William Bateson, one of Mendel's earliest and fiercest proponents, "introduced the terms ‘‘genetics’’, ‘‘homozygote’’, and ‘‘heterozygote’’ and emphasized that ‘‘fixity’’ of parental characteristics was of utmost importance in breeding. Bateson concluded that presentation by stating that ‘‘The period of confusion is passing away, and we have at length a basis from which to attack that mystery [heredity] such as we could scarcely have hoped two years ago would be discovered in our time." (This quote is from a 2008 tribute to Saunders.) (However, we should note that Bateson and many others did not think that these Mendealian traits, which were by definition faithfully transmitted and hence invariable -- as they then thought -- could have anything to do with evolution, which has to do with things that do vary).
Apparently Saunders quickly adopted Mendel's ideas, but Dr Comeau notes that he may have believed there was something Mendel had missed. Saunders wrote in 1910, "...I cannot help believing that the discovery of the supposed Mendelian units of inheritance may sometimes be due to the unfortunate combination of a lot of enthusiasm and of a very small number of factual observations." So, even early adopters of the idea of Mendelian inheritance were aware that it didn't explain everything.
But skip forward 100 years. Emphasis is still on Mendelian inheritance. The idea of "genes for" traits and diseases is still firmly held -- indeed, direct-to-consumer gene testing companies bank on it. Yes, there are Mendelian diseases, and thousands of causal genes have been identified but, in the intervening century, it seems that Mendelian traits have gotten more complex, while complex traits are still thought of as simple. That is, multiple, sometimes thousands of alleles have been identified for many Mendelian diseases (cystic fibrosis is the most obvious example), while the hope lives on that single genes are still to be found for diseases that are difficult to define and pretty clearly polygenic, and/or have a large environmental component (heart disease or type 2 diabetes or autism are examples of these).
The history in broad brush
During the past century, there were basically two branches of genetics. In one, Mendelian inheritance was used to find out much about genetic transmission and even to identify the nature of genes. Much of genetics was Mendelian, involved in segregation analysis which was the statistical attempt to show whether a given trait (and its presumed causal gene) was inherited as a Dominant, Recessive, Sex-Linked etc. manner. This formed the foundation of the very useful 'personalized' medicine called genetic counseling.
Fudge factors were introduced to accommodate the imperfect nature of the results -- uninherited instances were called 'phenocopies', and results that didn't fit the expected probabilities associated with diploid transmission (of one factor from a parent who carried two, as we do), and so on were introduced. These allowed the models to be fitted to the theory, without forcing adherents to the theory to take the issues very seriously. Partly, we didn't have the technical means to take it too seriously. But the nuances were noted!
The other branch of genetics was called quantitative genetics, relating to traits caused by many genes. Generally, these were quantitatively measured traits, like size, enzyme levels and so on. Polygenic traits were understood even in the early 1900s to be caused by many genes, even if we had almost no way to find them. Formal statistical theory, association and regression analysis basically, was developed by leading geneticists. They knew actual genes were involved, but didn't have the technical means to find them.
We were also aware of traits that seemed to cluster in families but not to segregate with Mendel's probabilities. So we compromised: they were threshold traits, based on a quantitative polygenic basis. Your genotype affected your blood pressure in a polygenic way, but if your blood pressure exceeded some limit, or threshold, you had a stroke. So the threshold value somewhat represented a quasi-Mendelian trait.
Gene mapping from the dawn of its earliest methods about 25-30 years ago brought everything full circle. The idea was that we could find the genes that 'caused' any old trait we were interested in. Big early 'hits' included cystic fibrosis, and some cancer-related genes such as the BRCA1 and BRCA2 genes. This led people to expect that everything could be predicted by identifying genes.
In turn, even when Mendelians themselves knew they had to introduce fudge factors, and quantitative geneticists knew that multiple genes contributed small effects, the public was sold on the idea that with Big Technology every trait would turn out to be enumerably 'genetic'.
Now in fact, these methods have shown that the 'simple' Mendelian traits' fudge factors reflected underlying complexity even at a single major causative gene. People who have mutations that are associated in others with severe diseases can show no sign of disease, for example. And polygenic traits turned out to be just that. But it turns out that the confusion between inheritance of genetic material, and the inheritance of traits, persists even when (unlike in Mendel's garden) we know better.
Thus, it is the full recognition of these realities that is being resisted by the Big Data community, who cling to the idea of simple answers with high predictive power. And that means, though Mendel should in a sense be consigned to a very honorable place in the history books of a science to which his work contributed so mightily, the tendency to exhume his ideas, to generate false promises, seems irresistible.
These principles were derived from Mendel's work with pea plants, a study of carefully chosen traits that behaved in a particular way that he felt would be useful for forming better agricultural hybrid plants -- when he crossed plants with different flower color or seed characteristics etc., the traits did not 'blend' in subsequent generations but instead remained discrete: the offspring had one flower color and one type of seed coat; green peas crossed with yellow ones didn't produce in-between colored peas. But, Mendel knew very well that not all traits behaved in this way; he just felt they weren't of interest for his purposes. And without actually saying so, he looked for simple patterns of inheritance that fit other 'hot' science of his time, the development of the atomic theory of matter.
In 1854, in preparation for his later work, Mendel acquired 34 varieties of pea seed, Pisum sativum, from local nurseries. He spent the next 2 years growing them to determine which characters bred true; that is, which parental traits would appear, unblended, in the offspring. Not all did. He chose 7 that did to follow through multiple generations. He produced 7 crosses, each repeated twice, from a seed plant with a particular trait and the second time with that same trait from the pollen plant. In this way he determined that the male and female contribute equally to the offspring.
He bred his plants for the next 8 years, counting the occurrence of the 7 traits in nearly 30,000 peas. Because he was astutely brilliant and chose his traits well, he was able to document what became known as recessive and dominant inheritance, as well as the random segregation of alleles in offspring, and the independent inheritance of genes for traits using a sample size large enough to produce convincing results. In essence he developed statistical rules for the apparent inheritance of his chosen traits. But, he knew he wasn't explaining all of inheritance. And later work on plants that didn't behave according to his laws further confirmed that he hadn't explained everything (sometimes to his frustration, or even despair, though he had moved on to monastery administration by then).
The story of how Mendel's work was largely ignored for 40 years, and rediscovered in 1900 is well-known (we blogged about it some time ago, too, here and here, e.g.). As genetics got under way in earnest in the next few decades, to a sufficient extent, given the technology of the time, Mendelian rules, indeed, seemed rather like 'laws' of nature, although it was clear they didn't explain everything.
![]() |
| Thomas Morgan in the fly room; Nobelprize.org |
One of the most influential and productive geneticists of the early era was Thomas Morgan. He and his students at Columbia created the first linkage maps in fruit flies, identifying chromosomal loci associated with traits in the 1910's, and so on. Even as they did so, Morgan was aware that most traits were due to many genes, and that it was likely that most genes do more than one thing. He wrote in 1917, e.g.:
A man may be tall because he has long legs, or because he has a long body, or both. Some of the genes may affect all parts, but other genes may affect one region more than another. The result is that the genetic situation is complex and, as yet, not unraveled. (The Theory of the Gene, p 294).And, Morgan ended the paragraph above with this sentence: "Added to this is the probability that the environment may also to some extent affect the end-product."
Morgan also knew that the genetic effect -- the 'trait'-- that he was studying could be restricted to just some developmental stage of his flies, but his point was not the trait but tracing its behavior relative to the idea that they had causal effects that were localized to particular places on chromosomes. The trait-complexity was something he explicitly said needed to wait for a later time when more was known.
A student of Morgan's, A.H. Sturtevant, in his A History of Genetics (1965), wrote:
With Johannsen [who introduced the words "gene", "genotype" and "phenotype" in the early 1900's] it became evident that inherited variations could be slight and environmentally produced ones could be large, and that only experiments could distinguish them.And:
In 1902 Bateson pointed out that it should be expected that many genes would influence such a character as stature, since it is so obviously dependent on many diverse and separately varying elements. This point of view was implied by Morgan in 1903 (Evolution and Adaptation, p. 277), and by Pearson in 1904.In the early 1900's extensive studies of even Mendel's traits showed variation not consistent with simple 2-factor control (Ken wrote an Evolutionary Anthropology piece discussing some of this). So, chromosomal loci associated with traits were being found, but it was also recognized that traits were complex.
We bring all this up because an MT reader, André Comeau, plant geneticist and pathologist, recently wrote to Ken with some thoughts on Canadian plant breeder (and flutist) Charles Saunders, developer of Marquis wheat, the most important variety of wheat of the last century in Canada. Saunders developed this wheat after a lot of cross-breeding of different wheats from around the world. Marquis wheat became commercially available in 1909 and because it had a shorter growing season than other available wheat varieties, by 1920, 90% of the wheat grown in Canada was Marquis.
In 1902, Saunders attended the lecture at which William Bateson, one of Mendel's earliest and fiercest proponents, "introduced the terms ‘‘genetics’’, ‘‘homozygote’’, and ‘‘heterozygote’’ and emphasized that ‘‘fixity’’ of parental characteristics was of utmost importance in breeding. Bateson concluded that presentation by stating that ‘‘The period of confusion is passing away, and we have at length a basis from which to attack that mystery [heredity] such as we could scarcely have hoped two years ago would be discovered in our time." (This quote is from a 2008 tribute to Saunders.) (However, we should note that Bateson and many others did not think that these Mendealian traits, which were by definition faithfully transmitted and hence invariable -- as they then thought -- could have anything to do with evolution, which has to do with things that do vary).
Apparently Saunders quickly adopted Mendel's ideas, but Dr Comeau notes that he may have believed there was something Mendel had missed. Saunders wrote in 1910, "...I cannot help believing that the discovery of the supposed Mendelian units of inheritance may sometimes be due to the unfortunate combination of a lot of enthusiasm and of a very small number of factual observations." So, even early adopters of the idea of Mendelian inheritance were aware that it didn't explain everything.
But skip forward 100 years. Emphasis is still on Mendelian inheritance. The idea of "genes for" traits and diseases is still firmly held -- indeed, direct-to-consumer gene testing companies bank on it. Yes, there are Mendelian diseases, and thousands of causal genes have been identified but, in the intervening century, it seems that Mendelian traits have gotten more complex, while complex traits are still thought of as simple. That is, multiple, sometimes thousands of alleles have been identified for many Mendelian diseases (cystic fibrosis is the most obvious example), while the hope lives on that single genes are still to be found for diseases that are difficult to define and pretty clearly polygenic, and/or have a large environmental component (heart disease or type 2 diabetes or autism are examples of these).
The history in broad brush
During the past century, there were basically two branches of genetics. In one, Mendelian inheritance was used to find out much about genetic transmission and even to identify the nature of genes. Much of genetics was Mendelian, involved in segregation analysis which was the statistical attempt to show whether a given trait (and its presumed causal gene) was inherited as a Dominant, Recessive, Sex-Linked etc. manner. This formed the foundation of the very useful 'personalized' medicine called genetic counseling.
Fudge factors were introduced to accommodate the imperfect nature of the results -- uninherited instances were called 'phenocopies', and results that didn't fit the expected probabilities associated with diploid transmission (of one factor from a parent who carried two, as we do), and so on were introduced. These allowed the models to be fitted to the theory, without forcing adherents to the theory to take the issues very seriously. Partly, we didn't have the technical means to take it too seriously. But the nuances were noted!
The other branch of genetics was called quantitative genetics, relating to traits caused by many genes. Generally, these were quantitatively measured traits, like size, enzyme levels and so on. Polygenic traits were understood even in the early 1900s to be caused by many genes, even if we had almost no way to find them. Formal statistical theory, association and regression analysis basically, was developed by leading geneticists. They knew actual genes were involved, but didn't have the technical means to find them.
We were also aware of traits that seemed to cluster in families but not to segregate with Mendel's probabilities. So we compromised: they were threshold traits, based on a quantitative polygenic basis. Your genotype affected your blood pressure in a polygenic way, but if your blood pressure exceeded some limit, or threshold, you had a stroke. So the threshold value somewhat represented a quasi-Mendelian trait.
Gene mapping from the dawn of its earliest methods about 25-30 years ago brought everything full circle. The idea was that we could find the genes that 'caused' any old trait we were interested in. Big early 'hits' included cystic fibrosis, and some cancer-related genes such as the BRCA1 and BRCA2 genes. This led people to expect that everything could be predicted by identifying genes.
In turn, even when Mendelians themselves knew they had to introduce fudge factors, and quantitative geneticists knew that multiple genes contributed small effects, the public was sold on the idea that with Big Technology every trait would turn out to be enumerably 'genetic'.
Now in fact, these methods have shown that the 'simple' Mendelian traits' fudge factors reflected underlying complexity even at a single major causative gene. People who have mutations that are associated in others with severe diseases can show no sign of disease, for example. And polygenic traits turned out to be just that. But it turns out that the confusion between inheritance of genetic material, and the inheritance of traits, persists even when (unlike in Mendel's garden) we know better.
Thus, it is the full recognition of these realities that is being resisted by the Big Data community, who cling to the idea of simple answers with high predictive power. And that means, though Mendel should in a sense be consigned to a very honorable place in the history books of a science to which his work contributed so mightily, the tendency to exhume his ideas, to generate false promises, seems irresistible.
Wednesday, December 11, 2013
Sometime geneticist Joe Terwilliger on genetics
We may recently have given the false impression that geneticist Joe Terwilliger gives less priority to science, or at least good science, than to other perhaps more frivolous pursuits (he is Abe Lincoln every February, for example, and tuba player the rest of the time -- unless he's cleaning up bean debacles as a diplomat, or being a basketball and language coach to Dennis Rodman), so we wanted to help correct any such misconceptions here. Perhaps to that end, Joe (now known in South Korea, we're afraid, as "sometime geneticist Joe Terwilliger") suggested we republish a blog post he first posted on his own short-lived blog in 2008. He recently dug this up again and says that few could disagree, even 5 years later.
The point is that sometimes there is a lot of convenient hard-of-hearing even in science, which fancies itself to be an objective search for truth. Some of the details in Joe's post are out-of-date but we, and he, think that the basic thrust is not. In a sense that makes the conclusion all the more cogent, because the same modes of thinking about genomic causation are still predominant, despite the vastly costly but essentially consistent results in the five years since 2008. And, as Joe points out, he and Ken had much the same message in 2000.
One not-so-subtle change, we will note, is that promises by NIH Director Francis Dr Collins, and many others in presumably responsible positions, have steadily altered their due date, which recedes into the distance like, say, an oasis as you grope for water, the fences if you want your pitchers to have a better earned run average, or a preacher's promises of ultimate salvation, if you weekly plunk coins into the basket.
So perhaps the lesson is that under these circumstances, rather than just dismiss critics, science--actual science as it's supposed to be--should feel a need to take stock of what it's doing. But we leave it to you to judge.
And if you hear about Joe in other contexts in weeks to come, remember that he was a sometime geneticist here first:
The Rise and Fall of Human Genetics and the Common Variant - Common Disease Hypothesis
By Joe Terwilliger
Nov 2008
There is an enormity of positive press coverage for the Human Genome Project and its successor, the HapMap Project, even though within the field the initial euphoric party when the first results came out has already done a full 180 to be replaced by the hangover that inevitably follows such excesses.
While in a truly bizarre propaganda piece, Francis Collins, in a parting sycophantic commentary (J Clin Invest. 2008 May;118(5):1590-605) claimed that the controversy about the CV/CD hypothesis was "... ultimately resolved by the remarkable success of the genetic association studies enabled by the HapMap project." He went on to list a massive table of "successful" studies, including loci for such traits as bipolar, Parkinson disease and schizophrenia, and of course the laughable success of ApoE and Alzheimer disease. To be objective about these claims, let me quote from what researchers studying those diseases had to say.
![]() |
| Joe as Abe on the balcony (but not of Ford's Theater) |
The point is that sometimes there is a lot of convenient hard-of-hearing even in science, which fancies itself to be an objective search for truth. Some of the details in Joe's post are out-of-date but we, and he, think that the basic thrust is not. In a sense that makes the conclusion all the more cogent, because the same modes of thinking about genomic causation are still predominant, despite the vastly costly but essentially consistent results in the five years since 2008. And, as Joe points out, he and Ken had much the same message in 2000.
One not-so-subtle change, we will note, is that promises by NIH Director Francis Dr Collins, and many others in presumably responsible positions, have steadily altered their due date, which recedes into the distance like, say, an oasis as you grope for water, the fences if you want your pitchers to have a better earned run average, or a preacher's promises of ultimate salvation, if you weekly plunk coins into the basket.
So perhaps the lesson is that under these circumstances, rather than just dismiss critics, science--actual science as it's supposed to be--should feel a need to take stock of what it's doing. But we leave it to you to judge.
And if you hear about Joe in other contexts in weeks to come, remember that he was a sometime geneticist here first:
The Rise and Fall of Human Genetics and the Common Variant - Common Disease Hypothesis
By Joe Terwilliger
Nov 2008
There is an enormity of positive press coverage for the Human Genome Project and its successor, the HapMap Project, even though within the field the initial euphoric party when the first results came out has already done a full 180 to be replaced by the hangover that inevitably follows such excesses.
For
those of you not familiar with the history of this field and the controversies
about its prognosis which were present from the outset, I refer you to a review
paper I and a colleague wrote back in 2000 at the height of the controversy - Nature Genetics 26, 151 - 157 . The basic gist
of the argument put forward for the HapMap project was the so-called common
variant/common disease hypothesis (CV/CD) which proposed that "most of the
genetic risk for common, complex diseases is due to disease loci where there is
one common variant (or a small number of them)" [Hum Molec Genet
11:2417-23]. Under those circumstances it was widely argued that using the
technologies being developed for the HapMap project, that one would be able to
identify these genes using "genome-wide association studies" (GWAS),
basically by scoring the genotype for each individual in a cross sectional
study for each of 500,000 to 1,000,000 individual marker loci - the argument
being that if common variants explained a large fraction of the attributable
risk for a given disease, that one could identify them by comparing allele
frequencies at nearby common variants in affected vs unaffected individuals.
This point was contested by researchers only with regard to how many markers
you might have to study for this to work if that model of the true state of
nature applied. Many overly optimistic scientists initially proposed 30,000
such loci would be sufficient, and when Kruglyak suggested it might take
500,000 such markers people attacked his models, yet today the current
technological platforms use 1,000,000 and more markers, with products in the
pipelines to increase this even more, because it quickly became clear that the
earlier models of regular and predictable levels of linkage disequiblrium were
not realistic, something that should have been clear from even the most basic
understanding of population genetics, or even empirical data from lower
organisms.
Today
such studies are widespread, having been conducted for virtually every disease
under the sun, and yet the number of common variants with appreciable
attributable fractions that have been identified is miniscule. Scientists have
trumpetted such results as have been found for Crohn's disease, in which 32
genes were detected using panels of thousands of individuals genotyped at
hundreds of thousands of markers - this sounds great until you start looking at
the fine print, in which it is pointed out that all of these loci put together
explain less than 10% of the attributable risk of disease, and for various
well-known statistical reasons, this is a gross overestimate of the actual
percentage of the variance explained. Most of these loci individually explain
far less than half a percent of the risk, meaning that while this may be
biologically interesting, it has no impact at all on public health as most of
the risk remains unexplained. This is completely opposite to the CV/CD theory
proposed as defined above. In fact, this is about the best case for any complex
trait studied, with virtually every example dataset I have personally looked at
there is absolutely nothing discovered at all.
At
the beginning of the euphoria for such association studies, the example
"poster child" used to justify the proposal was the relationship
between variation at the ApoE gene and risk of Alzheimer disease. In an
impressively gutsy paper recently, a GWAS study was performed in Alzheimer
disease and published as an important result, with a title that sent me rolling
on the floor in tears laughing: "A high-density whole-genome association
study reveals that APOE is the major susceptibility gene for sporadic
late-onset Alzheimer's disease" [ J Clin Psychiatry. 2007 Apr;68(4):613-8
] - in an amazingly negative study they did not even have the expected number
of false positive findings - just ApoE and absolutely nothing else... And the
authors went on to describe how important this result was and claimed this
means they need more money to do bigger studies to find the rest of the genes.
Has anyone ever heard of stopping rules, that maybe there aren't any common
variants of high attributable fraction??? This was a claim that Ken Weiss and I
put forward many times over the past 15 years, and Ken has been making this
point for a decade before that even, in his book, "Genetic variation and human
disease", which anyone working in this field should read if they are not
familiar with the basic evolutionary theory and empirical data which show why
noone should ever have expected the CV/CD hypothesis to hold...
In
many other fields, the studies that have been done at enormous expense have
found absolutely nothing, and in what Ken Weiss calls a form of Western Zen (in
which no means yes), the failure of one's research to find anything means they
should get more money to do bigger studies, since obviously there are things to
find but they did not have big enough studies with enough patients or enough
markers - it could not possibly be that their hypotheses are wrong, and should
be rejected... It is a truly bizarre world where failure is rewarded with more
money - but when it comes to promising upper-middle-aged men (i.e. Congress)
that they might not die if they fund our projects, they are happy to invest in
things that have pretty much now been proven not to work...
While in a truly bizarre propaganda piece, Francis Collins, in a parting sycophantic commentary (J Clin Invest. 2008 May;118(5):1590-605) claimed that the controversy about the CV/CD hypothesis was "... ultimately resolved by the remarkable success of the genetic association studies enabled by the HapMap project." He went on to list a massive table of "successful" studies, including loci for such traits as bipolar, Parkinson disease and schizophrenia, and of course the laughable success of ApoE and Alzheimer disease. To be objective about these claims, let me quote from what researchers studying those diseases had to say.
Parkinson
disease: "Taken together, studies appear to provide substantial evidence
that none of the SNPs originally featured as PD loci (sic from GWAS studies)
are convincingly replicated and that all may be false positives...it is worth
examining the implications for GWAS in general." Am J Hum Genet 78:1081-82
Schizophrenia:
"...data do not provide evidence for involvement of any genomic region
with schizophrenia detectable with moderate [sic 1500 people!] sample
size" Mol Psych 13:570-84
Bipolar
AND Schizophrenia: "There has been great anticipation in the world of
psychaitric research over the past year, with the community awaiting the
results of a number of GWAS's... Similar pictures emerged for both disorders -
no strong replications across studies, no candidates with strong effect on
disease risk, and no clear replications of genes implicated by candidate gene
studies." - Report of the World Congress of Psychiatric Genetics.
Ischaemic
stroke: "We produced more than 200 million genotypes...Preliminary
analysis of these data did not reveal any single locus conferring a large effect
on risk for ischaemic stroke." Lancet Neurol. 2007 May;6(5):383-4.
And
the list goes on and on of traits for which nothing was found, with the authors
concluding they need more money for bigger studies with more markers. It is
really scary that people are never willing to let go of hypotheses that did not
pan out. Clearly CV/CD is not a reasonable model for complex traits. Even the
diseases where they claim enormous success are not fitting with the model -
they get very small p-values for associations that confer relative risks of
1.03 or so - not "the majority of the risk" as the CV/CD hypothesis
proposed.
One
must recall that in the intial paper proposing GWAS by Risch and Merikangas
(Science 1996 Sep 13;273(5281):1516-7) - a paper which, incidentally, pointed
out that one always has more power for such studies when collecting families
rather than unrelated individuals - the authors stated that "despite the
small magnitude of such (sic: common variants in)genes, the magnitude of their
attributable risk (the proportion of people affected due to them) may be large
because they are quite frequent in the population (sic: meaning >>10% in
their models), making them of public health significance." The obvious
corollary of this is that if they are not quite frequency, they are NOT having
high attributable fraction and are therefore NOT of public health significance.
And
yet, you still have scientists claiming that the results of these studies will
lead to a scenario in which "we will say to you, 'suppose you have a 65%
chance of getting prostate cancer when you're 65. If you start taking these
pills when you're 45, that percent will change to 2". Amazing claims when
the empirical evidence is clear that the majority of the risk of the majority
of complex diseases is not explained by anything common across ethnicities, or
common in populations... (Leroy Hood, quoted in the Seattle
Post-Intelligencer). Francis Collins recently claimed that by 2020, "new
gene-based designer drugs will be developed for ... ALzheimer disease,
schizophrenia and many other conditions", and by 2010, "predictive
genetic tests will be available for as many as a dozen common conditions".
This does not jibe with the empirical evidence... In Breast Cancer for example,
researchers claimed that knowledge of the BRCA1 and BRCA2 genes (which confer
enormously high risk of breast cancer to carriers) was uninteresting as it had
such a small attributable fraction in the population. Of course now they have
performed GWAS studies and examined tens of thousands of individuals and have
identified several additional loci which put together have a much smaller
attributable fraction than BRCA1 and BRCA2, yet they claim this proves how
important GWAS is. Interesting how the arguments change to fit the data, and
everything is made to sound as if it were consistent with the theory.
I
suggest that people go back and read "How many diseases does it take to
map a gene with SNPs?" (2000) 26, 151 - 157. There are virtually no
arguments we made in that controversial commentary 8 years ago which we could
not make even stronger today, as the empirical data which has come up since
then basically supports our theory almost perfectly, and refutes conclusively
the CV/CD hypothesis, despite Francis Collins' rather odd claims to the
contrary...
In
the end, these projects will likely continue to be funded for another 5 or 10
years before people start realizing the boy has been crying wolf for a damned
long time... This is a real problem for science in America, however, as NIH is
spending big money on these rather non-scientific technologically-driven
hypothesis-free projects at the expense of investigator-initiated
hypothesis-driven science. Even more tragically training grants are enormously
plentiful meaning that we are training an enormous number of students and
postdocs in a field for which there will never be job opportunities for them,
even if things are successful. Hypothesis-free science should never be allowed
to result in Ph.D. degrees if one believes that science is about questioning
what truth is and asking questions about nature, while engineering is about how
to accomplish a definable task (like sequencing the genome quickly and
cheaply). The mythological "financial crisis" at NIH is really more a
function of the enormous amounts of money going into projects that are
predetermined to be funded by political appointees and government bureaucrats
rather than the marketplace of ideas through investigator-initiated proposals.
Enormous amounts of government funding into small numbers of projects is a bad
idea - one which began with Eric Lander's group at MIT proposing to build large
factories for the sequencing of the genome rather than spreading it across
sites, with the goal of getting it done faster (an engineering goal) instead of
getting more sites involved so that perhaps better scientific research could
have come along the way. This has led to a scenario years later in which the
factories now want to do science and not just engineering, which is totally
contrary to their raison d'etre, and leads to further concentrations of funding
in small numbers of hands when science is better served, perhaps by a larger
number of groups receiving a smaller amount of money so that more brains are
working in different directions thinking of novel and innovative ideas not
reliant on pure throughput. Human genetics has transformed from a field with
low funding, driven by creative thinking into a field driven by big money and
sheep following whatever shepherd du jour is telling them they should do (i.e.
innovative means doing what they current trend is rather than something truly
original and creative). This is bad for science, and also is bad science. GWAS
has been successful technologically, and it has resoundingly rejected the CV/CD
hypothesis through empirical data. If we accept this and move on, we can put
the HapMap and HGP where it belongs, in the same scientific fate as the
Supercollider, and let us get back to thinking instead of throwing money at
problems that are fundamentally biological and not technological!
(most
notably in terms of the big money NIH is sending into these non-scientific
technologically-driven hypothesis-free studies, rather than investigator
initiated hypothesis-driven science - one of the main causes of the
"funding crisis" at NIH where a tiny portion of new grants are funded
- get rid of the big science that is not working - like the supercollider! -
and there is no funding crisis)
Tuesday, February 12, 2013
Concentric circles of causation
Public health has always been a broad discipline, encompassing the study of diseases in populations in all its forms, applied and academic both, and it's getting broader. The story of the first epidemiologist, John Snow, is well-known -- he suspected that an 1854 cholera epidemic on Broad Street in London was due to a waterborne infectious agent, at a time when it was generally thought that cholera was an airborne disease. He convinced authorities to remove the handle from the pump on the well that he suspected was the source of the epidemic. It turned out, of course, that he was right, although the spread of cholera had already slowed by the time the handle was taken off the pump. So, his contribution was to future public health, not his ill contemporaries.
Epidemiology is the study of patterns of disease in populations -- who's at risk, what's the cause, how to prevent it. For more than 100 years after John Snow epidemiologists concentrated on proximate single causes or specific exposures -- cigarettes and lung cancer, Legionnaire's disease and Legionella bacteria, HIV and AIDS -- but recently the field of "social epidemiology" has gained some traction. This is the study of the social determinants of health and disease. The proximate cause of AIDS is HIV, but often, HIV is contracted through drug use and shared needles, or patterns of multiple sex partners. And, in most places, those at highest risk are poor, so that it's perfectly legitimate to say that poverty causes AIDS. So, what's the actual 'cause' of the disease? And, where does Public Health intervene to control it?
There has long been debate over whether epidemiology has a theoretical framework, or whether it is just a set of established statistical methods. Social epidemiologist Nancy Krieger, in her seminal paper in 1994 called "Epidemiology and the web of causation: has anyone seen the spider?" discussed just this, writing that epidemiology at the time, yes, was interested in the 'web of disease causation' -- what causes disease in populations? -- but was neglecting the search for the spider, the maker of the web. This paper kick-started the field of social epidemiology.
Krieger wrote:
There's a new field in biology, too; 'systems biology'. This is the study of interconnected biological systems -- metabolic pathways, gene interactions, cell signaling networks. This is intended to be a more holistic approach to biology rather than a reductionist one. The idea is that describing these interactions will help us to understand the 'emergent' traits that are complex diseases. It is hoped that there will be practical applications -- understanding networks will, e.g., elucidate 'druggable' pathways that pharmaceutical companies can then intervene on, to prevent or cure disease.
Ken and I have often criticized the idea that the new field of 'systems biology' is going to be all it promises. Like social epidemiology, systems biology is a somewhat self-congratulatory jargonized term whose proponents suggest the very reasonable idea of taking a broader view of causation, in an attempt to take into account as many factors as possible that might explain how genes function or how traits are made or the causes of a disease. It's basically a way of enumerating identifiable interactions among components.
So, rather than being satisfied with knowing that mutations in a specific gene on their own cause a particular disease, the idea is to take a larger view: Enumerate the regulatory pathways or gene interactions (that is, between proteins) so that there are more targets for pharmaceutical intervention.
Systems biology sounds like a good idea. A standard, purely additive model of what genes do treats each gene like an independent dose of some effect, and the overall result is just the sum of these effects. But this is clearly oversimplifying at best, because gene products interact in various molecular ways that need not just be additive. Instead, sets of genes (their coded proteins, or the proteins that cause a given gene to be expressed) interact with each other -- in 'systems'.
Metabolism is an example in which chains or cycles or hierarchies of interactions pass molecules from one stage to the next. Even if the interaction networks are the mechanism of action, there is no reason to expect that the effects of variation in the components will lead to simple or additive variation in the result.
However, the temptation (whether explicit or implicit) is to try to rescue complexity by treating networks as self-contained causal units. Take sets of tens or hundreds of genes and treat each as a single causal unit, and magically you've reduced the causal dimensions you need to consider by an order of magnitude! But this is wishful thinking, and doesn't really simplify causation or the hunt to understanding, because we know that genes contribute to multiple networks, that vary, overlap, and have alternative pathways that are used under different circumstances.
Social epidemiology faces similar complexity.
Broad causal networks may seem to smooth out causal relationships until they look generalizable, but they aren't necessarily any smoother in social than molecular life. There's a common issue in population sciences called the "ecological fallacy." This is when an association that is true on the population level is inferred to be true on the individual level. The county may always vote Republican, or this neighborhood may be a wealthy one, but it can't be assumed that everyone in the county votes Republican, or that everyone in the neighborhood has an income above the average. Or that the neighborhood causes political preference or wealth in a given individual.
Social epidemiology by its nature and objectives searches for population level associations and attempts to infer from those associations causal relationships that apply at the level of the individual. Poverty causes AIDS. Racism causes stress which causes high blood pressure. In a sense, as reasonable and plausible as these ideas, and as much as they must reflect causal processes in some way, the ecological fallacy is always lurking, because not everyone with AIDS is poor, and not everyone who is poor gets AIDS, not everyone who has experienced racism has high blood pressure, and so on. And, even eliminating poverty won't prevent further cases of AIDS.
Unlike systems biology, where the idea is to intervene in the networks with targeted pharmacological agents, it's hard to know what to do with the information that social epidemiology is producing. Public Health is, ultimately, an applied science. It has very little that could be called real theory, beyond the use of statistical methods to design and evaluate studies -- that is, assuming the kinds of repeatabliity that any statistical sampling requires, similar to saying that every roll of dice has the same probability of coming up 6. Dice rolls may be repeatable events, but to a great extent, humans aren't. And as with systems biology, the idea is not just to enumerate interactions but to identify targets of intervention. But, when the field is identifying "poverty" or "racism" as causal factors, what can be done to intervene?
Indeed, one might say that at least simple biological systems like the krebs energy cycle or photosynthesis require networks of interactions among a step-wise hierarchy of truly enumerable molecular components in quantitative relationships, and a very high degree of repeatability so they can be studied experimentally and evaluated with standard statistical methods. But are 'poverty' and 'racism' even serious concepts of similar type? If not, then current statistical or other study-design methods may simply be inappropriate, too vague, or unable to provide the type of answers we would like to get, especially if we desire to understand causation at the individual level.
Public Health, and epidemiology in particular, has a legacy of successful research and intervention in infectious diseases -- these are 'point source' diseases, with largely replicable causality, and if you cut the problem off at the source, you prevent the disease. This has its parallels with human genetics, where clearly Mendelian, single-gene disorders are relatively easy to explain. It's when causation gets complex that genetics -- and Public Health -- get into difficulties.

But, ok, granted, Public Health is a population-level field, and eliminating all cases of a disease has never been asked of it. But it's curious to see the field seeming to homogenize causation at a time when complexity is a buzzword in other fields. Though, again, this is quite in line with much of genetics, which has its own legacy of successes with point causation (Mendelian disease), just not so much with complex diseases. And which also mixes population-level effects of particular variants with the ability of those variants to 'personalize' medical care.
Have you seen the photos that Col. Chris Hadfield is posting on his Facebook page? He's in the Space Station, taking pictures of Earth in his free time and sharing them with the world. They are stunning. He photographs cities, regions, large chunks of continents; they are lit up at night or green, or snow-covered, or cloud-covered, or dry-as-a-bone desert, or stretches of ocean during the day. When he wonders what something is in one of his pictures, he crowdsources the answer. The other day it was a spot in Tehran that looked curious to him. It turned out it had once been an airport, and is now a playground, which Iranians told him.
If a region is experiencing drought, that's easy to see from these photos. But, if there are cracks in the macadam on those long stretches of highway, or a bridge is unstable, or the rivers are polluted, you can't see it from space. At least, not from these pictures. You do get the big picture, but you have to zoom in to discover where to intervene to prevent a catastrophe due to crumbling infrastructure.
Social epidemiology may be a lot like Col. Hadfield's photographs -- beautiful descriptions, but so far removed from the individual that it's impossible to actually predict who's going to get sick and why, never mind useful for figuring out where to intervene. Even identifying the risk factors themselves is notoriously difficult. And one need not give it a special name, which is often a way we academics have of making their ideas seem new or particularly insightful. Just 'epidemiology' will do very well--and the point is to include social as well as other potentially causal factors (and, as we hope is clear, we are not just picking on public health rhetoric, because genetics is just as much affected by unnecessary jargonizing).
Public Health is by goals and design a population-based field, and it works well when the message is that cholera is waterborne, so we need to keep our water sources clean, or that vaccination can protect entire populations against infectious disease. Its methods have saved countless lives. But, like trying to make risk predictions from genetic data from whole populations, it's hard to know what to do with these descriptions of the causes of complex diseases from such a distance.
And ultimately, of course, if we go out enough concentric circles of causation, it's life that is the cause of disease, and the only prevention is death.
Epidemiology is the study of patterns of disease in populations -- who's at risk, what's the cause, how to prevent it. For more than 100 years after John Snow epidemiologists concentrated on proximate single causes or specific exposures -- cigarettes and lung cancer, Legionnaire's disease and Legionella bacteria, HIV and AIDS -- but recently the field of "social epidemiology" has gained some traction. This is the study of the social determinants of health and disease. The proximate cause of AIDS is HIV, but often, HIV is contracted through drug use and shared needles, or patterns of multiple sex partners. And, in most places, those at highest risk are poor, so that it's perfectly legitimate to say that poverty causes AIDS. So, what's the actual 'cause' of the disease? And, where does Public Health intervene to control it?
There has long been debate over whether epidemiology has a theoretical framework, or whether it is just a set of established statistical methods. Social epidemiologist Nancy Krieger, in her seminal paper in 1994 called "Epidemiology and the web of causation: has anyone seen the spider?" discussed just this, writing that epidemiology at the time, yes, was interested in the 'web of disease causation' -- what causes disease in populations? -- but was neglecting the search for the spider, the maker of the web. This paper kick-started the field of social epidemiology.
Krieger wrote:
[This paper] emphasizes why epidemiologists must look first and foremost to the link between social divisions and disease to understand etiology and to improve the public’s health, and in doing so exposes the incomplete and biased slant of epidemiologic theories reliant upon a biomedical and individualistic world-view.She's just published a new paper in the American Journal of Public Health, "History, Biology and Health Inequities: Emergent Embodied Phenotypes and the Illustrative Case of the Breast Cancer Estrogen Receptor" in which she argues that health inequities can only be reduced if diseases are considered not just as static biological entities, but within their social and evolutionary contexts, and as individual histories (she calls this broad view of, in this case the estrogen receptor, the "emergent embodied phenotype" -- we can ignore the jargonizing of what is a rather obvious idea). The paper was brought to our attention by Susan Oyama, who has done some very good work on developmental systems, broadening the understanding of genes in context, among other things.
There's a new field in biology, too; 'systems biology'. This is the study of interconnected biological systems -- metabolic pathways, gene interactions, cell signaling networks. This is intended to be a more holistic approach to biology rather than a reductionist one. The idea is that describing these interactions will help us to understand the 'emergent' traits that are complex diseases. It is hoped that there will be practical applications -- understanding networks will, e.g., elucidate 'druggable' pathways that pharmaceutical companies can then intervene on, to prevent or cure disease.
Ken and I have often criticized the idea that the new field of 'systems biology' is going to be all it promises. Like social epidemiology, systems biology is a somewhat self-congratulatory jargonized term whose proponents suggest the very reasonable idea of taking a broader view of causation, in an attempt to take into account as many factors as possible that might explain how genes function or how traits are made or the causes of a disease. It's basically a way of enumerating identifiable interactions among components.
So, rather than being satisfied with knowing that mutations in a specific gene on their own cause a particular disease, the idea is to take a larger view: Enumerate the regulatory pathways or gene interactions (that is, between proteins) so that there are more targets for pharmaceutical intervention.
![]() |
| Seoul from space |
Metabolism is an example in which chains or cycles or hierarchies of interactions pass molecules from one stage to the next. Even if the interaction networks are the mechanism of action, there is no reason to expect that the effects of variation in the components will lead to simple or additive variation in the result.
However, the temptation (whether explicit or implicit) is to try to rescue complexity by treating networks as self-contained causal units. Take sets of tens or hundreds of genes and treat each as a single causal unit, and magically you've reduced the causal dimensions you need to consider by an order of magnitude! But this is wishful thinking, and doesn't really simplify causation or the hunt to understanding, because we know that genes contribute to multiple networks, that vary, overlap, and have alternative pathways that are used under different circumstances.
Social epidemiology faces similar complexity.
Broad causal networks may seem to smooth out causal relationships until they look generalizable, but they aren't necessarily any smoother in social than molecular life. There's a common issue in population sciences called the "ecological fallacy." This is when an association that is true on the population level is inferred to be true on the individual level. The county may always vote Republican, or this neighborhood may be a wealthy one, but it can't be assumed that everyone in the county votes Republican, or that everyone in the neighborhood has an income above the average. Or that the neighborhood causes political preference or wealth in a given individual.
Social epidemiology by its nature and objectives searches for population level associations and attempts to infer from those associations causal relationships that apply at the level of the individual. Poverty causes AIDS. Racism causes stress which causes high blood pressure. In a sense, as reasonable and plausible as these ideas, and as much as they must reflect causal processes in some way, the ecological fallacy is always lurking, because not everyone with AIDS is poor, and not everyone who is poor gets AIDS, not everyone who has experienced racism has high blood pressure, and so on. And, even eliminating poverty won't prevent further cases of AIDS.
Unlike systems biology, where the idea is to intervene in the networks with targeted pharmacological agents, it's hard to know what to do with the information that social epidemiology is producing. Public Health is, ultimately, an applied science. It has very little that could be called real theory, beyond the use of statistical methods to design and evaluate studies -- that is, assuming the kinds of repeatabliity that any statistical sampling requires, similar to saying that every roll of dice has the same probability of coming up 6. Dice rolls may be repeatable events, but to a great extent, humans aren't. And as with systems biology, the idea is not just to enumerate interactions but to identify targets of intervention. But, when the field is identifying "poverty" or "racism" as causal factors, what can be done to intervene?
Indeed, one might say that at least simple biological systems like the krebs energy cycle or photosynthesis require networks of interactions among a step-wise hierarchy of truly enumerable molecular components in quantitative relationships, and a very high degree of repeatability so they can be studied experimentally and evaluated with standard statistical methods. But are 'poverty' and 'racism' even serious concepts of similar type? If not, then current statistical or other study-design methods may simply be inappropriate, too vague, or unable to provide the type of answers we would like to get, especially if we desire to understand causation at the individual level.
Public Health, and epidemiology in particular, has a legacy of successful research and intervention in infectious diseases -- these are 'point source' diseases, with largely replicable causality, and if you cut the problem off at the source, you prevent the disease. This has its parallels with human genetics, where clearly Mendelian, single-gene disorders are relatively easy to explain. It's when causation gets complex that genetics -- and Public Health -- get into difficulties.

But, ok, granted, Public Health is a population-level field, and eliminating all cases of a disease has never been asked of it. But it's curious to see the field seeming to homogenize causation at a time when complexity is a buzzword in other fields. Though, again, this is quite in line with much of genetics, which has its own legacy of successes with point causation (Mendelian disease), just not so much with complex diseases. And which also mixes population-level effects of particular variants with the ability of those variants to 'personalize' medical care.
Have you seen the photos that Col. Chris Hadfield is posting on his Facebook page? He's in the Space Station, taking pictures of Earth in his free time and sharing them with the world. They are stunning. He photographs cities, regions, large chunks of continents; they are lit up at night or green, or snow-covered, or cloud-covered, or dry-as-a-bone desert, or stretches of ocean during the day. When he wonders what something is in one of his pictures, he crowdsources the answer. The other day it was a spot in Tehran that looked curious to him. It turned out it had once been an airport, and is now a playground, which Iranians told him.
If a region is experiencing drought, that's easy to see from these photos. But, if there are cracks in the macadam on those long stretches of highway, or a bridge is unstable, or the rivers are polluted, you can't see it from space. At least, not from these pictures. You do get the big picture, but you have to zoom in to discover where to intervene to prevent a catastrophe due to crumbling infrastructure.
Social epidemiology may be a lot like Col. Hadfield's photographs -- beautiful descriptions, but so far removed from the individual that it's impossible to actually predict who's going to get sick and why, never mind useful for figuring out where to intervene. Even identifying the risk factors themselves is notoriously difficult. And one need not give it a special name, which is often a way we academics have of making their ideas seem new or particularly insightful. Just 'epidemiology' will do very well--and the point is to include social as well as other potentially causal factors (and, as we hope is clear, we are not just picking on public health rhetoric, because genetics is just as much affected by unnecessary jargonizing).
Public Health is by goals and design a population-based field, and it works well when the message is that cholera is waterborne, so we need to keep our water sources clean, or that vaccination can protect entire populations against infectious disease. Its methods have saved countless lives. But, like trying to make risk predictions from genetic data from whole populations, it's hard to know what to do with these descriptions of the causes of complex diseases from such a distance.
And ultimately, of course, if we go out enough concentric circles of causation, it's life that is the cause of disease, and the only prevention is death.
Subscribe to:
Posts (Atom)


.jpg)







