17. February 2012 · Comments Off on Are loss-of-function variants relevant to complex disease? · Categories: Papers · Tags: , , , , , , , ,

Three members of the group (James, Luke & Jeff) were involved in a substantial undertaking (led by our colleague Daniel MacArthur) to study loss-of-function (LoF) variants in otherwise “normal” human genomes. These are mutations which are predicted to obliterate the function of a gene: things like gained stop codons, coding frameshift insertions and the like. The paper has just come out in Science, and Daniel has a great write-up over at Genomes Unzipped. We made two contributions to the project (reflecting our principal interests):

  1. Because real LoF variants are (generally) selected against, but sequencing errors which look like LoF aren’t, this class of variation is hugely enriched in all sorts of sequencing and annotation errors. We therefore spent a lot of effort in hand validating LoF calls (using Evoker) to try to separate the wheat from the chaff.
  2. One might expect LoF variants to be more likely than an average SNP to affect disease risk, but it didn’t really seem to be the case. Only one very well known LoF variant (the Crohn’s disease NOD2 frameshift) from the paper showed any appreciable association to disease. Perhaps these variants are just too strongly selected against to rise to frequencies visible to GWAS?