Saturday, October 20, 2012

The Sum of All Genetics

A interesting paper was published a few days back on the genetics of autism that suggested that the genetic component of autism is made up of not one or two large mutations but rather many small mutations that add together to increase risk.

As the paper put it -
For simplex families, who have only a single affected individual in multiple generations, approximately 40% of liability traces to additive effects whereas this narrow-sense heritability exceeds 60% for ASD individuals from multiplex families.
Or to put that another way, you can only blame genetic mutations for roughly half the risk of autism.  This idea isn't necessarily new but it is the first time that I have seen it presented so openly in a paper with the names of so many prominent autism researchers attached.

I'm not going to dig into the methods used in the paper because, quite frankly, I have no real understanding of the methods used to arrive at the result and so I don't have an opinion on whether the results fit the data.  If you want a good summary of the results in the paper, I suggest you read the write up over at Questioning Answers.

What I am going to point out is the excellent background section at the start of the paper.  The paper is open access, so I if you are interested in the genetic of autism I suggest you read at least that section.  I also wanted to point out two little bits from this section -
Beliefs about the genetic architecture of autism spectrum disorders (ASD) have changed dramatically over the past few decades. Early twin studies produced heritability estimates approaching 90% and, while no specific risk loci were known at the time, it was believed that liability was conferred by a handful of genes of large effect. Later, data on the distribution of ASD within families, together with results from linkage analyses, were interpreted to mean that liability arose from many genes. Recent work has definitively demonstrated the substantial contribution of de novo variation. Indeed multiple studies of rare single nucleotide and copy number variants (CNVs) have suggested that 15% or more of liability traces to de novo mutation, effects that are genetic but not inherited.
[D]espite a near-consensus that common and transmitted variation must confer liability, multiple genome-wide association studies have so far not revealed replicable common polymorphisms associated with ASD, and studies of rare structural and sequence mutations have largely failed to account for the anticipated risk associated with transmitted variation.
I may be reading too much into these statements, but it almost seems like the paper is admitting that the assumption that autism is primarily an inherited genetic disorder is flawed.  Of course, the data in the paper tries to save the idea by suggesting that the genetic risk for autism comes from tens, hundreds, or thousands of small inherited mutations which, when acting together, give a greater risk for autism.

And who knows, that idea may actually be true.  Maybe the genetic side of autism does in fact come from many small mutations acting together.  But here is the thing about science - it is all about the evidence.  An estimate from applying some fancy algorithm to an already over analyzed genetic data set doesn't tell us anything new.

Despite a "near-consensus" that there have to be common genetic mutations that are passed from parent to child that increase the risk of autism, the fact is that after years upon years of looking no one has found them.  All that has been found are some relatively rare inherited conditions and a bunch of non-inherited mutations that, when taken together, make up the minority of cases of autism.

In fact, I would say that the lack of findings despite the efforts to date make a pretty convincing case that such common mutations do not exist, the evidence of twin and other studies notwithstanding.  Maybe it is time to start looking in more promising areas such as epigenetics or the biology of autism?


Klei L, Sanders SJ, Murtha MT, Hus V, Lowe JK, Willsey AJ, Moreno-De-Luca D, Yu TW, Fombonne E, Geschwind D, Grice DE, Ledbetter DH, Lord C, Mane SM, Lese Martin C, Martin DM, Morrow EM, Walsh CA, Melhem NM, Chaste P, Sutcliffe JS, State MW, Cook EH Jr, Roeder K, Devlin B. Common genetic variants, acting additively, are a major source of risk for autism. Mol Autism. 2012 Oct 15;3(1):9. [Epub ahead of print] PubMed PMID: 23067556 DOI : 10.1186/2040-2392-3-9


  1. Hi there, I'm a PhD student doing research in the field of psychiatric genetics and just thought I would make a couple of points (having just read the paper you discuss) as I feel I have some understanding of what these authors did.

    1-Although the analysis in this paper does contain some simulations (or as you put it "fancy algorithms"), the majority of the analysis is not based on simulation but uses the Genome-wide Complex Trait Analysis (GCTA) (, a relatively new tool, to compare how related ASD probands are to each other across the whole genome relative to how related controls are to each other across the genome (with separate comparisons between 1st degree relatives and controls). I can't pretend I understand the ins and outs of the method but I know this method has been used on other data sets successfully and it seems like a very clever technique of estimating heritability from common variants.
    2- I just attended the World Congress of Psychiatric Genetics last week (a large international scientific conference) where I got the strong impression that the consensus in the field of psychiatric genetics is that the sample sizes so far have been too small to detect common variants in many disorders, particularly in childhood conditions. Indeed, the sample sizes for ASD for genome-wide analyses so far, although they may seem like they're large (being in the thousands), are relatively tiny when compared to those that have been analysed in schizophrenia (tens of thousands), which have now detected over 60 common SNPs (single genetic variants) that are significantly involved in the condition. Genetically, there are a lot of parallels between these 2 devastating disorders but actually research from other genetic conditions/traits that are unrelated to either ASD or schizophrenia show the same trend in terms of how many samples are needed in the analysis to find true association of common variants with these complex traits/conditions (e.g. height or Alzheimer's disease).

    The point I am trying to make is that I think it is too early to dismiss the contribution of "common" variants (which lie on a wide spectrum of how common they actually are, as all things are relative). That said, there are numerous other worthwhile research avenues worth pursuing at the same time, including de novo rare variants, epigenetics, early environmental problems etc.
    I hope that makes sense and helps a little bit in understanding the issues brought up by this complex paper!

    1. Hi Joana,

      Thanks for the comment. I understand what you are saying and partially agree with you. Genetics are still do play a role in autism and common variants might still be important.

      But I think that the genetic side of autism, whatever it might be, has received more than its fair share of the attention. While researchers spend their time searching for small genetic changes that might account for some small part of the autism risk, there are other areas of autism that are much more promising that are being completely ignored.

      On a somewhat genetic level, there are epigenetic mechanisms and programming that haven't received that much attention even though there is some reason to think that the problems seen in autism could be cause by this sort of change.

      And then there are biological aspects of autism and their effects that are almost completely ignored. A good example here is sleep in children with autism. Almost half of children with autism have sleep issues and these issues (presumably) have something to do with disruption of how melatonin is produced and used.

      If you wanted to make a major difference in the lives of children with autism, you could spend some time tracking down exactly why people with autism have extremely low levels of melatonin and don't show the typical night/day production patterns.

      That's just one simple example, there are many other biological findings that are extremely common in autism, from immune disruption to low levels of essential substances such as zinc and cholesterol. All of these things appear in large parts of the autism population, or at the very least far more than 10% of the population

      So, since there are limited research dollars available, it would seem that we could get more bang for the buck to look into some of these other areas. It might even make more sense if autism is in fact caused by hundreds of little mutations acting together. It seems like it would be easier to find disrupted pathways and track backwards to find the problem than it would be to try and piece together hundreds of unrelated pieces into a picture of what could conceivably be disrupted in people with autism.