Common SNPs, Rare CNVs... and the Expression Network Between?

Background:  Recent genetic studies in common, complex heritable disease like autism spectrum disorders have met success in two arenas: geneticists like to divide differences among individuals into ‘common polymorphisms’ and ‘rare variants’, and often argue about which kind of variation will be more important in disease risk.  Both kinds of results provide challenges to direct translation into neurobiological understanding of autism, as they rarely reveal easily interpretable protein-coding mutations.  Common polymorphism association signals often fall in noncoding sequences or between genes, and rare variants in autism have primarily been large deletions and duplications including many brain-expressed genes. 

Objectives:  I will describe an approach to ‘pathway’ definition using gene expression data that can unite a number of autism-implicated common polymorphisms and rare variants into a network with functional implications.

Methods:  In a previous genome wide association study (GWAS), significant association with autism was detected near the SEMA5A gene, which has also shown evidence for reduced expression in autism (Weiss et al, 2009).  Here we have used public expression and genetic data in controls to define eQTLs and master regulators for SEMA5A expression in lymphoblast cell lines.  We have gone on to test for SNP association and CNV association in autism datasets within these putative expression networks using set-based approaches.  Further, we have used RNAi knock-down techniques in lymphoblast cell lines to functionally validate the genetic-expression networks we have identified.

Results:  We have identified SNP association in one large autism GWAS dataset.  We have identified CNV association in 4 autism datasets.  We also show cellular expression data for human lymphoblast cell lines.

Conclusions:  Our approach of defining an expression regulatory pathway for a SNP-associated candidate gene has revealed additional common, and now rare, variants associated with autism and may provide a framework for identifying which rare CNVs are likely to contribute to autism risk.