ASHG: Epistasis and the Missing Heritability

As if no time has passed at all since the sunny shores and lost laptops of the American Society of Genetics 2009 meeting, ASHG2010 has rolled around, this time in Washington DC. As always, I’m going to be trying to write a few thoughts on the conference every day, though this year it may be split between here and Genomes Unzipped.

I’ll also be semi-live-tweeting (wifi coverage is patchy), so you can get up-to-the-minute details of all the talks on my twitter feed (@lukejostins), or from other tweeps via the hashtag #ASHG2010.

Epistasis and Missing Heritability

As Daniel observed on Twitter, I very nearly had a heart attack when Eric Lander, in his Distinguished Speaker’s talk about the Human Genome Project, said that the “missing heritability” is probably all down to Epistasis (i.e. interactions between variants). His argument was that GWAS had low power to detect gene-gene interactions, and therefore there could be lots hanging around that count account for the unexplained variance.

This is a fallacy, and a big one. The power to detect interaction between any two variants is indeed very low, but the power to assess the effect of interactions as a class is very high; instead of inferring lots of interactions, you can merely infer a single parameter, the expected degree of interaction. Eric’s statement is like saying that because you cannot see a tree from a mile away, you won’t know that there is a forest there.

People have indeed set out to infer the degree of gene-gene interaction in complex disease. In a PLoS Genetics paper a few years back David Clayton demonstrated that there is actually very high power to detect deviation from simple multiplicity. He showed that there is evidence for some interactions in Type I Diabetes, but that these effects are almost entirely driven by the very strong HLA signal. If we look at normal variants, there is very little interaction at all.

We can go one better. David Clayton’s paper calculated the predictive power (in the form of the AUC) of a logistic regression model with and without interaction terms for type 1 diabetes. Naomi Wray recently published a paper showing how you can convert between the AUC and the proportion of heritability explained, and, even better, produced an online calculator for doing exactly that.

Plugging David AUCs into Naomi’s calculator tells us that the non-HLA variants explain 10% of the heritability of Type I Diabetes. If we include interactions, this increases to 11% of the heritability. So, while epistasis exists, and increases the proportion of heritability explained, the increase is nowhere near enough to account for the “missing heritability”.

Share and Enjoy:
  • Digg
  • Reddit
  • StumbleUpon
  • Facebook
  • Twitter
  • Google Bookmarks
  • FriendFeed

4 Responses to ASHG: Epistasis and the Missing Heritability

  1. Are you placing any bets on what will explain it?

  2. Hmm. Being somewhat flakey, but I expect it’ll really fall apart under a number of gradual effects. 3% more will come from better fine mapping and another 3% from secondary effects at known loci, both of which will fall out with greater sample size. Perhaps another 10% lives in very low-effect common variation, and another 5% from epistasis. Another 10% could live in low-frequency variation of medium effect (OR 2 or so), and then maybe another 5% in rare variation. I recon we’ll also see a drop in the estimates of heritability. I’d also guess that a chunk of it is effectively unfindable (massively polygenic, rare but low-effect, that sort of thing).

  3. What about the effects due to most diseases likely being phenotypically similar but genotypically heterogeneous ? I would not call that “massively polygenic” as poorly characterized.

  4. I don’t know if epistasis is the answer or not — presumably all will be revealed when Lander et al. publish their eagerly anticipated paper. In the meantime, there’s an interesting letter to the editor in the European Journal of Human Genetics from Harvard’s David Haig (2011) 19, 123;
    doi:10.1038/ejhg.2010.161;) who independently supports the epistasis idea. And I quote:

    “A conceptual difference between pedigree studies and GWAS does
    not appear to have been considered: pedigree-based heritability measures
    the phenotypic effects of much larger chunks of chromosome than
    GWAS-based heritability. This distinction can be illustrated with a
    simple example that elides complexities arising from diploidy. Consider
    two SNPS (A/T and G/C) in linkage equilibrium that are located 0.1 cM
    apart. The SNPs could, for example, encode two amino acid substitutions
    within a single protein. From the perspective of pedigree-based measures
    of heritability, the four haplotypes (AG, AC, TG and TC) are inherited as
    four alleles at a single locus, but from the perspective of GWAS these are
    biallelic polymorphisms at distinct loci. Suppose that the combinations
    AG and TC add a little bit extra to height but AC and TG subtract a little
    bit. Then, neither SNP will be correlated with height in GWAS, but the
    haplotypes, which are correlated with height, will be reliably transmitted
    from parents to offspring and will contribute to estimates of pedigree-based
    heritability. Put another way, the genetic effect on phenotype
    appears as part of the additive genetic variance in pedigree studies but
    as an unmeasured gene-gene interaction in GWAS.
    “The major constraint on measuring interactions in GWAS has been the
    very large number of possible interactions. If there are 10E6 SNPs on an
    array, then there are 5X10E11 pairs of SNPs. However, the number of pairs is a much more manageable 10E6 if analysis is restricted to neighboring SNPs.”

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>