Elucidating the Effects of Genetic Variants On Protein Coding and Noncoding Genes in Autism

Background: The past few years have witnessed substantial advances in understanding the genetic basis of autism. These advances have been made possible by high-throughput genotyping and genome-wide association analysis for identification of genetic variants associated with autism. This approach has revealed single nucleotide polymorphisms (SNPs), copy number variants (CNVs), and microinserts and deletions (InDels) providing insights into the genetic basis of autism. However, the full breadth of the goals of the Autism Genome Project and the emergence of high-throughput genotyping for genome-wide association studies are rapidly running into several bottlenecks in translating genomic findings into clinical practice to improve human health. One of the more significant bottlenecks is the inability to elucidate the effects of genetic variants (SNPs, CNVs and InDels) on protein coding and noncoding genes. While GWAS has powerfully identified genetic variants, majority of these variants (>60%) are located in noncoding and intergenic regions, and much work remains to elucidate their effects on gene function.
Objectives: The objective of this study is to elucidate the effects of genetic variants (SNPs, CNVs, InDels) on protein coding and noncoding genes.

Methods: We have and continue to use the power of computational systems biology framework to (a) create an inventory of the genetic variants that contribute to autism risk and (b) to leverage genotype data with sequence data to elucidate the effects of identified genetic variants on protein coding and noncoding genes.

Results:   We have performed computational analysis of the effects of genetic variants on cis regulatory elements, regulatory regions and splice sites in 260 autism candidate genes currently in the Autism Candidate Gene Map (ACGMap) Database that we have created. In more than 60% of the genes we have shown that genetic variants, particularly structural variants (CNVs and InDels) can adversely affect or disrupt cis regulatory elements, regulatory regions and splicing events. Furthermore, through ab initio prediction, we have identified potential regions for targeted sequencing.

Conclusions: We show that the effects of genetic variants on gene function can be elucidated computationally. Given the prevalence of genetic variants in noncoding regions and the significance of these regions with respect to gene regulation, it is imperative that cis regulatory elements disrupted by these variants be identified. While experimental methods are essential for validating predicted cis regulatory elements, computational prediction provides a quick and cost effective approach.