Friday, March 15, 2013

Functional illiteracy and genetic background

Doesn't everyone know this?
It's rather astonishing that such a paper still has to be written, but it does.  A nice chip in the misleading but still widely assumed one gene = one phenotype paradigm appeared online February 28 in Trends in Genetics.  The paper, "Does your gene need a background check? How genetic background impacts the analysis of mutations, genes, and evolution," Chandler, Chari and Dworkin, addresses the "vast complexity [that] has been omitted from our current understanding of allelic affects."  That is, genetic background, "the context-dependent nature of genetic effects."
We assert that a broad understanding of genetic effects and the evolutionary dynamics of alleles requires identifying how mutational outcomes depend upon the ‘wild type’ genetic background. Furthermore, we discuss how best to exploit genetic background effects to broaden genetic research programs.
While genetic determinism, the idea that traits are determined by genes, primarily single genes, and that there are genes 'for' most traits is still widespread, even among biologists -- yep, we rail against it all the time here on MT and elsewhere -- complexity hasn't been entirely ignored.  It has been known for some time, for example, that many alleles have strain-dependent effects in inbred mice, or many other species, for that matter, perhaps even lethal in one strain, with no effect in others.  And, Mendelian traits in humans, predominantly influenced by a single gene, can vary widely among the affected population.  Further, epistasis, the effect of  'modifier genes' on the expression of other genes, has long been a working concept in genetics.  And so on. But, there's a difference between what people know in theory, and what they put into practice (as they say, if you can read but don't, it's as good as being illiterate).

Here are two examples of background effect from the paper.  The 'wild type' fruit fly wing and mutant alleles in different strains in the figure is on the left, a wild type C. elegans tail and the effects of a mutant allele on the right. The single mutation has different effects in two different strains in both species, that is, when introduced into strains with different genetic backgrounds.



Figure 2. Induced mutations often have qualitatively or quantitatively variable effects on organismal phenotypes in different genetic backgrounds and in different environments. These effects can range from mild (in some cases perhaps even resulting in phenotypes that are indistinguishable from the wild type) to severe. (a) The scallopedE3 allele has qualitatively distinct effects on wing morphology in two commonly used wild type strains of Drosophila melanogaster, despite the wild type wings being qualitatively similar across these backgrounds. These background effects extend to include epistatic interactions between sd and other loci [1]. (b) The effects of the tra-2(ar221); xol-1(y9) genotype on sexual differentiation in the tail of Caenorhabditis elegans vary quantitatively with both rearing temperature and wild type genetic background [2]. The effects of genetic background are most apparent at intermediate temperatures. Source: TIGs, Chandler et al., Feb 28, 2013
 
The authors argue that the effect of genetic background is too often dismissed as noise, but that it should be used to advantage for understanding basic genetics as well as evolution. Epistatic effects, they suggest, are surely frequently responsible for contradictory results across strains, but further, can have "significant clinical and economic impact."

An example of the effective illiteracy that comes of ignoring complexity offered by the authors is the story of fruit fly experiments that identified a gene 'for' longevity, the I'm not dead yet (Indy) gene.  Flies with a particular Indy allele lived a lot longer than flies without.  Until the mutation was introduced into another strain of fly.  Clearly, the allele interacts with other genes, and/or environmental factors, to extend lifespan, but this is something researchers don't always consider when choosing fly strains, and interpretation of results can be controversial. 

Further, Chandler et al. write, the evolutionary fate of a genetic mutation may well depend on the genetic background in which it occurs.  And, if a background modifier allele has no effect on fitness, and only spreads by drift, "stochastic forces play a role in determining which regions of 'genotype space' are accessible to populations."  Or, if the mutation is pleiotropic, with effects on traits that are themselves influenced by selection, which helps them spread, it might then have an effect on the spread of traits that are themselves selectively neutral.

The authors offer further cautions; most traits vary continuously, but researchers 'discretize' them for convenience.  This can bias interpretation of results, since it's assumed that dependence of the expression of an allele on genetic background is controlled "by one or more modifiers of major effect" when in fact it could be influenced by variation in many genes. Environmental and other factors can influence allelic expression, and their effect, too, can depend on genetic background.  And so forth.

And of course the effect of a mutation can vary within strains as well -- inbred strains, where the assumption is that the effect should be the same in all individuals.  That it often doesn't is presumably because developmental processes are inherently random to some degree, but this introduces a whole new wrinkle to interpretation of experimental results, though it is often overlooked.  Researchers often will publish only the most dramatic or extreme result, without mentioning that results in fact varied.

Chandler et al. also point out that "the background-dependent phenotypic effect may not reflect the interaction of the background with the lesion per se, but may instead reveal more about other genetic processes, such as the molecular machinery influencing RNAi or the somatic effects of transposable elements (TEs) on gene expression."

More (and less)
So far so good.  But let's step out of the lab, away from inbred organisms and into real life.  In that context, saying that the effect of allele A depends on genetic 'background' has to be thought about more seriously.  Taken at face value, and applied to disease genetics, that implies that if we just have large enough samples the background can be considered just statistical noise, as its effects will 'even out' in large samples, revealing the 'true' effect of allele A. But properly conceived, statistical 'noise' refers to things like measurement errors and extraneous variables, which is not the case at all here!  Invoking the hope of being salvaged by large numbers is a mistake, or, rather is an assumption that we know is simply false.  It's the assumption that allele A acts alone, or additively independent of other aspects of the genome.

Examples in which the same allele in this strain of flies does one thing, but a different thing in another strain or species, are used as in Chandler to make this point.  But we have to be careful not to think that the background effect is due to one allele at one other gene, which Chandler et al. point out.  Typically, it will be due to some combination of variants at many other genes, and with recombination and mutation occurring every generation in every individual, not to mention varying environments, we can't be sure just which other genes or combinations of variants at which genes, are responsible for the difference.

There are many reasons why allele A doesn't act alone.  But even if it did, the average effect of allele A doesn't really say much about what the allele does, or its predictive value in an individual (but the latter is what we are being promised in the hyped 'personalized genomic medicine' and direct-to-consumer snake oil senses).  In some backgrounds Allele A may kill you, in others it doesn't do anything (there are legions of examples of this), but without replication of the same allele in the same sets of backgrounds (which is impossible since each person's background is unique), we usually simply can't say which backgrounds these harmful ones are with much clarity--we just know we've seen fatal phenotypes in individuals with the allele.  Predicting who it will kill is another matter.

Many of us carry numerous dead genes (that is, genes with inactivating alleles in them) without discernible harm, and it's been estimated that about 10% of truly pathologic alleles in humans are the normal allele in other vertebrates.  For example, the apoE e4 allele that is so often claimed to be a  'cause' of Alzheimer's disease, is the normal allele in our close primate relatives, with no evidence we know of that aging primates are highly susceptible to dementia.  Background matters.

Clear thinking is too rare, and in part because it is too honest for the grant system that we have
Functional illiteracy is a problem.  When the underlying assumption is genes for traits, this can impede interpretation and understanding of genotype/phenotype relationships.  Everybody knows this.  Now it's time for everyone to act as if they know it.

No comments: