Samocha et al. A framework for the interpretation of de novo mutation in human disease. Nat Genet. 2014 Aug 3.
Samocha et al describe a sophisticated statistical model designed to better evaluate data derived from mass exome sequencing studies, specifically with respect to the significance of excesses of de novo mutations in diseases with significant locus heterogeneity such as autism spectrum disorders. They derive the predicted number of de novo mutations for a given gene by combining several parametres such as gene length, sequence context, depth of coverage during sequencing, and regional divergence around the gene between humans and macaques. This allows a comparison between expected and observed de novo mutations for groups of affected and unaffected individuals, which yields a more powerful analysis than direct comparison between groups. Likewise, the model allows the identification of groups of genes that, in normal subjects, exhibit less de novo functional variation than expected, implying a selective pressure against loss of function (“constrained genes”).
The authors then apply this framework to the study of autism spectrum disorders. They find that there is no global excess of de novo functional mutation in ASD subjects, but rather an excess of genes with more than one de novo functional mutation, as well as an excess of de novo mutations in “constrained” genes, as well as in genes that are the targets of FMRP. Surprisingly, these observations were not true of ASD subjects with IQ above 100, suggesting that the genetic underpinnings of ASD are different in this subgroup.
The analysis of exome data for multifactorial diseases exhibiting significant locus heterogeneity requires different, and more complex, bioinformatic tools than for monogenic disorders. This study is an example of the value of developing and perfecting them.