RT Journal Article
SR Electronic
T1 Functional annotation of rare structural variation in the human brain
JF bioRxiv
FD Cold Spring Harbor Laboratory
SP 711754
DO 10.1101/711754
A1 Lide Han
A1 Xuefang Zhao
A1 Mary Lauren Benton
A1 Thaneer Perumal
A1 Ryan L. Collins
A1 Gabriel E. Hoffman
A1 Jessica S. Johnson
A1 Laura Sloofman
A1 Harold Z. Wang
A1 CommonMind Consortium
A1 Kristen J. Brennand
A1 Harrison Brand
A1 Solveig K. Sieberts
A1 Stefano Marenco
A1 Mette A. Peters
A1 Barbara K. Lipska
A1 Panos Roussos
A1 John A. Capra
A1 Michael Talkowski
A1 Douglas M. Ruderfer
YR 2019
UL http://biorxiv.org/content/early/2019/07/23/711754.1.abstract
AB Structural variants (SVs) contribute substantially to risk of many brain related disorders including autism and schizophrenia. However, annotating the potential contribution of SVs to disease remains a major challenge. Here, we integrated high resolution SV calling from genome-sequencing in 755 human post-mortem brains with dorsal lateral prefrontal cortex RNA-sequencing from a subset of 629 samples to quantify the dosage and regulatory effects of SVs. We show that genic (p = 5.44×10−9) and regulatory SVs (enhancer p = 3.22×10−23, CTCF p = 3.86×10−18) are present at significantly lower frequencies than intergenic SVs after correcting for SV length. Copy number variants (CNVs)—deletions and duplications—exhibit a significant quantitative and directional relationship between the proportion of genic and regulatory content altered and gene expression, and the size of the effect is inversely correlated with the loss-of-function intolerance of the gene. We trained a joint linear model that leverages genic and regulatory annotations to predict expression effects of rare CNVs in independent samples (R2 = 0.21-0.41). We further developed a regulatory disruption score for each CNV that aggregates the predicted expression across all affected genes weighted by the genes’ intolerance score and applied it to an independent set of SVs from 14,891 genome-sequenced individuals. Pathogenic deletions implicated in neurodevelopmental disorders by ClinGen had significantly more extreme regulatory disruption scores than the rest of the SVs. Rank ordering based on the most extreme regulatory disruption scores prioritized pathogenic deletions that would not have been prioritized by frequency or length alone. This work points to the deleteriousness of regulatory SVs, particularly those altering CTCF sites. We further provide a simple approach for functionally annotating the regulatory effects of SVs in the human brain that has potential to be useful in larger SV studies and should improve as more regulatory annotation data is generated.