RT Journal Article
SR Electronic
T1 Statistical model integrating interactions into genotype-phenotype association mapping: an application to reveal 3D-genetic basis underlying Autism
JF bioRxiv
FD Cold Spring Harbor Laboratory
SP 2020.07.27.222364
DO 10.1101/2020.07.27.222364
A1 Qing Li
A1 Chen Cao
A1 Deshan Perera
A1 Jingni He
A1 Xingyu Chen
A1 Feeha Azeem
A1 Aaron Howe
A1 Billie Au
A1 Jun Yan
A1 Quan Long
YR 2020
UL http://biorxiv.org/content/early/2020/07/27/2020.07.27.222364.abstract
AB Biological interactions are prevalent in the functioning organisms. Correspondingly, statistical geneticists developed various models to identify genetic interactions through genotype-phenotype association mapping. The current standard protocols in practice test single variants or single regions (that contain multiple local variants) sequentially along the genome, followed by functional annotations that involve various aspects including interactions. The testing of genetic interactions upfront is rare in practice due to the burden of testing a huge number of combinations, which lead to the multiple-test problem and the risk of overfitting. In this work, we developed interaction-integrated linear mixed model (ILMM), a novel model that integrates a priori knowledge into linear mixed models. ILMM enables statistical integration of genetic interactions upfront and overcomes the problems associated with combination searching.Three dimensional (3D) genomic interactions assessed by Hi-C experiments have led to unprecedented biological discoveries. However, the contribution of 3D genomic interactions to the genetic basis of complex diseases has yet to be quantified. Using 3D interacting regions as a priori information, we conducted both simulations and real data analysis to test ILMM. By applying ILMM to whole genome sequencing data for Autism Spectrum Disorders, or ASD (MSSNG) and transcriptome sequencing data (GTEx), we revealed the 3D-genetic basis of ASD and 3D-eQTLs for a substantial proportion of gene expression in brain tissues. Moreover, we have revealed a potential mechanism involving distal regulation between FOXP2 and DNMT3A conferring the risk of ASD.Software is freely available in our GitHub: https://github.com/theLongLab/Jawamix5Competing Interest StatementThe authors have declared no competing interest.