FUNCTIONAL CHARACTERIZATION OF GENETIC LIABILITY FOR AUTISM SPECTRUM DISORDER

Our understanding of the genetic basis of autism spectrum disorder (ASD) is advancing rapidly. The number of ASD-associated, linkage disequilibrium independent risk loci is constantly growing due to ongoing PGC efforts. The majority of the common variants reside within non-coding regions of the genome and, as such, the formulation of testable hypotheses to elucidate their potential function in the brain is challenging. Integration of genetic findings with large-scale genotype-tissue brain expression efforts have enabled the execution of independent transcriptome-wide association studies (TWAS); however, heritability mediated by gene expression as assayed in homogenate tissues with RNA-seq is only estimated to account for about 10% of the total heritability. Recent advances in cell type specific transcriptomic and epigenetic profiling of the human brain are highlighting the unrealized potential of integrating cell-type specific transcriptomic and epigenetic features, such as chromatin accessibility and histone modifications, towards elucidating the mechanisms of cis-regulation of brain gene expression and causal gene prioritization. Here, we are leveraging large-scale high resolution multi-omics datasets derived from brain homogenate, FANS/FACS-sorted cells/nuclei and single-nucleus data to increase our mechanistic understanding of the genetic liability for autism spectrum disorder as follows: 1) We prioritize putatively causal SNPs associated with changes in transcriptomes and chromatin accessibility. 2) We quantify heritability mediated by brain gene expression. 3) We perform a transcriptome-wide association study to identify ASD-associated transcripts and biological pathways. 4) We perform epigenome-based prioritization of relevant cell types and tissues. 5) We identify cell subtypes exhibiting high expression across ASD-associated genes. Finally, we are exploring associations of aggregated genetic liability for ASD across complex cognition domains by leveraging several genotyped cohorts.