SSC-ASC Whole-Genome Sequencing Consortium (project 2): Development of statistical methods

The last decade has seen remarkable progress in gene discovery for autism spectrum disorder (ASD). This has been catalyzed by exome sequencing of the 1.5 percent of the genome that encodes proteins and microarray analysis to find very large copy number variations (CNVs). The findings are mainly driven by highly deleterious de novo mutations in the genes most intolerant to such variation. The next great challenge is whole-genome sequencing (WGS) to assess how regulatory variants in the rest of the genome contribute to ASD, and what insights these may give into when and where in the brain gene disruptions can cause symptoms.

WGS analysis is complicated by the sheer quantity of de novo and inherited variants across the genome and the limited tools available to reliably predict functional consequences. The scale and complexity of WGS-based studies thus require expertise in human genetics, genome biology, computer science, bioinformatics, statistical genetics and functional modeling. Based on this need for complementarity, the Simons Simplex Collection (SSC)-Autism Sequencing Consortium (ASC) Whole-Genome Sequencing Consortium was established to analyze data from the SSC and ASC cohorts. In a series of four linked projects, this consortium will expand on its initial efforts to aggregate, analyze and interpret WGS studies in ASD¹, as well as evaluating the effects of associated variants on gene function.

In project 2 of this consortium, Bernie Devlin and Kathryn Roeder aim to increase the power of association tests for WGS data. Devlin and Roeder will build a more powerful test for annotations by using their correlation structure; develop a de novo score, guided by the nature of the data (e.g., rarity) and will use high-dimensional inference techniques such as lasso and spectral analysis to select variation for the score; and determine which annotations or classes of annotations are significantly associated with ASD, using a new area of statistics called post-selection inference. Results from this work, performed within the SSC-ASC Whole-Genome Sequencing Consortium, will be freely shared with the community.