Monday, January 28, 2013

Genomic analysis results: understanding, or Fairy Dust?

We are daily seeing claims of major discoveries from genomics and other 'omics' kinds of studies.  These are being proclaimed by their investigators as if they have waved a magic wand and solved critical human problems 'urgently' in need of solution.

Yet, many realize that GWAS and other omics, or idea-free methods, have provided a much lower  yield than was promised or expected.  This is expressed in terms like 'hidden heritability', referring to the familial clustering that should be genetic but for which specific genes cannot be found, or at least many individually trivial contributing genome regions are identified.  In fact, this is what we should have expected, based on long-standing evolutionary theory and ideas about genetics.  We've posted many, many times about this.

The evidence is consistent.  Many genes interact to produce biological traits, in humans as well as other species including yeast and bacteria, and plants.  These genes have to be regulated to control the timing and amount of their expression in cells, and gene regulation involves many interactions among genes and other DNA regions where regulatory proteins bind.  Each of the functional DNA regions that are involved is subject to mutation that, if not lethal, can circulate in the population over generations.

This is known as 'polygenic' variation.  The word simply means a great many contributing genetic elements that mainly have individually tiny effects.  Findings from GWAS and other types of studies consistently point to evidence for just this kind of polygenic control.  But the frustrating thing (for proponents of genetics-are-everything and of personalized genomic medicine, etc.) is that with many individually trivial contributions, each person's genotype is different and each case of the 'same' disease is due to different genotypes and/or environmental exposures.

At the same time, major mutational changes in contributing genes can yield a serious effect that proper analysis can assign to that specific gene.  Our methods identify these, and we generally refer to their effects as 'Mendelian'.  These are often due to changes that inactivate ('knock out') the gene.  This success in easily identifying causes large and only problematically finding the small ones suggests that the reason there appears to be so much genetic control (reflected in measures like family correlation or heritability) is simply what we think it is: traits really are polygenic.

But can life be that complicated??
In the face of this apparent complexity--many argue that life can't really be that complex.  One may feel that it's just not plausible that hundreds or thousands of genes can be the explanation for traits that show orderly value distributions in populations.  That orderliness, and the relatively orderly nature of evolution, and the fact that a trait can be knocked out by single genes, all might be seen as indicating that life must have been able to evolve our complex traits in a way that is not so complex after all.  We're just not understanding--yet!

The usual approach to this view is to argue that we just need longer, larger, costlier studies, or more kinds of 'omics' approaches--like epigenetics, copy number variation, nutrigenomic, microbiomic studies and the like.  Then, this view goes, we'll (whew!) finally identify essentially all the causal elements.  But if things really are polygenic, this may be hopeless.

But, do we really know what's going on, whether or not causation is totally enumerable as the current belief system holds?  We know this belief system is based in part--perhaps major part--on the kinds of professional vested interests and paucity of better ideas that we write often about here.  One way to view this is imply to assert, yet again, that for nearly a century we have had the right kinds of knowledge and the right interpretation, even if lacking in sufficient technology to document those ideas, and that recent technologies are showing just what we expected to find--despite the resistance to the contrary, the idea that we can reduce complexity to simple genetics, or even omics, is largely based on wishful thinking.

But what about an alternative?

Fairy Dust?
Source: Wikimedia
Suppose this is one of those situations in which we are documenting the hell out of trivia, because our theoretical understanding leads us there--we try to force the current 'paradigm' to fit facts that really don't fit!  But perhaps not only are the wishful-thinkers wrong, but so are those of us who have been arguing that we see what we expected to see and that, alas, life really is as complex as our polygene theory suggests it is.

If both sides of these issues are wrong, perhaps there is some other explanation for what seems like the tractable theory of life's coherence, something other than many tiny contributing factors.  Could there be some force or factor--call it 'Fairy Dust'--that we simply have not discovered but that underlies what we are struggling to understand?

Such factors would be analogous to those that were discovered in other 'paradigm shifts', or revolutionary changes in scientific gestalt, that have happened over time.  We might refer to quantum effects, dark energy or dark matter, gravity particles, and so on as exemplars of such factors in other sciences. It could be some kind of 'field' or 'force' whose nature we don't know of or even suspect.  Or just another way of thinking about what we know already.

One can never deny the possibility that Fairy Dust exists.  But neither can we propose studies to find it, as if we knew it existed, and it's understandable given human nature and the history of science that we'll press ahead, ever more intently, trying herd-like to force things to fit our theory, or trying to outwit our theory, the way that's par for the course now, until someone somehow stumbles on the insight required to identify the fairy dust and improve our biological explanations.  This is just how Thomas Kuhn described the way science works (as we posted about yesterday).

But don't hold your breath, because to us, right now, it does not seem that our explanations are missing any such thing.

5 comments:

insolemexumbra said...

I look forward to you guys winning the Nobel for your discovery of Fairy Dust. ;)

Seriously though, well argued.

Ken Weiss said...

Thanks (but we won't share the Nobel winnings with you just because you said this!)

It is all too tempting to think one sees deeper into Nature than anybody else does or has. At the same time, I think Kuhn's general idea about 'normal' science and genuine 'paradigm shifts' have some cogency.

Here, maybe we are all accepting standard materialism (ignoring the unsuspected Fairy Dust), and arguing about whether complexity is due to what it seems to be due to, or how to understand it--but all on both sides arguing from within the same Darwinian/Wright-Fisher/evolutionary genetic paradigm.

Suppose some sort of 'field' (interactions among molecules in the cell, dark matter or energy, or some other factor totally unknown to us) exists. Once it is found, many things could fall into place.

Then, after the herd stampeded to research that new finding, we would see why we had the differences we had over polygenic complexity, and why we didn't get much closer to the truth that way.

Or, of course, maybe we DO understand things but too many just don't want to face up to it.

Or, maybe Enlightenment-era science, based on reductionism and replicability and 'laws' of Nature is just not suited for the kind of non-replicability that to a great extent is what life on earth is about.

John R. Vokey said...

There is always the possibility that Enlightenment-era science is/was NOT based on reductionism and replicability and 'laws' of Nature. Yes, it pushed empiricism as the rightful antidote to authority, but that is not *all* it did. Indeed, it is always refreshing to re-read Popper and his ``Conjectures and refutations'' lecture to see the core of the Enlightenment in plain English. There is no question he would condemn precisely what you describe as it is explanation-free "science"; i.e., not science at all. And, before him, Dewey, James, Peirce, and Holmes. If you haven't read The Metaphysical Club http://www.amazon.ca/Metaphysical-Club-L-Menand/dp/0374528497/ref=sr_1_1?s=books&ie=UTF8&qid=1359441688&sr=1-1 , I highly recommend it.

Ken Weiss said...

These things are always oversimplified, and of course I'm not an historian. In a sense, the digested net message from that era, as I see it, is what I was referring to.

There were debates about (and including) Darwin and whether he used proper scientific method in developing evolution, and there were long debates within cultural anthropology about how far just gathering data could take one.

Not only that, but most of us in the academic world put in our caveats, see nuance, are inconsistent, and so on.

I don't know Popper's paper but will try to get it. The person I've not read any of but keep reading about is Peirce. The Metaphysical Club sounds familiar, but I don't remember if I've read it---showing how well I do/don't remember things!

Anyway, thanks much for these suggestions. But my point in this post, at least, wasn't a critique of the Enlightenment (nor an oversimplification if I committed that fault), but to raise the kind of "Kuhnian" issue of whether we are pressing up against the limits of our current 'paradigm' or whether the latter really is basically correct but much in the research community doesn't want to face it because it doesn't generate simple truths.

I think we never can know when someone may really discover the next Fairy Dust, however, showing how we've all been wrong, and that was what we tried to say.

Again, your comments are always thoughtful and helpful, and thanks for offering them!

Ken Weiss said...

John,
If you are referring to Popper's paper and not the whole book of the same title, I've now read that paper.

I am not sure what points you were making in citing him, other than the observation that pure induction leads nowhere without some deductive-type of generalization derived from (or held before)the application of repeated observations (induction).

But that paper is not about reductionism or other relevant issues, so perhaps I'm not reading what you had in mind.

In any case, Popper's idea of falsification hasn't fared that much better over the years, despite its widely held if misperceived invocation.

To me, 'omics' research is exactly what Popper was criticizing: it is almost pure induction, based on the belief--and that's the right word for it, because 'axiom' would glorify its shallowness) that genes (or gene expression, or nutrition, or whatever 'omics' one is advocating) simply must be responsible. So the hypothesis tests are in a sense fakes, and I would say the idea of hypothesis-free work is also a fake for the same reason. One assumes what one believes and one can always find it in data--that, certainly, was a major point of Popper in the paper I read.

I think this is relevant to our post today, about how we know what we think we know. Popper was pre-Kuhn and to my knowledge didn't address Kuhn's sociological analysis (maybe he did--I'm not well read in Popper, though Kuhn was anticipated in the 1930's by Fleck and Popper probably did read Fleck).

Today's omics has some successes as anything so systematically and technically well done will. But in a way it is a very different way to do science that I think really does harken back to more purely 'Baconian' ideas that observation leads to theory. Anyway, this is debatable and too much for blog-post commentary!