Shownotes Transcript
Link to bioRxiv paper: http://biorxiv.org/cgi/content/short/2023.03.02.530529v1?rss=1
Authors: Wu, A. P.-Y., Singh, R., Walsh, C. A., Berger, B.
Abstract: Genome-wide association studies (GWAS) often report disease-linked genetic variation at noncoding genomic loci, but it is difficult to identify which genes these loci affect. Advances in single-cell multimodal assays that profile chromatin accessibility and gene expression in the same cell hold promise for addressing this challenge, as they can reveal causal locus-gene relationships. However, existing computational approaches for estimating such relationships from single-cell data are unable to account for what we refer to as "cell-state parallax," the time lag between epigenetic and transcriptional modalities due to their cause-and-effect relationship. In dynamic biological processes, chromatin regulatory potential is a manifestation of this parallax which we hypothesized could be recovered from single-cell snapshots. Leveraging the econometric concept of predictive temporal causality, we newly identify tissue-specific locus-gene associations where accessibility of the locus predicts expression of the gene. Our algorithm, GrID-Net, is a neural network-based generalization of classical Granger causal inference that enables graph-structured analysis, instead of being limited to sequential orderings. We applied it to a single-cell study of human corticogenesis to predict neuronal cis-regulatory elements, enabling us to interpret genetic variants in schizophrenia (SCZ). We identified 132 genes linked to common noncoding SCZ variants, including the potassium transporters KCNG2 and SLC12A6, and we present a strategy for unveiling the regulatory mechanisms underlying gene dysregulation. Our work points to the transformative potential of single-cell multimodal assays for discovering novel gene regulatory mechanisms and provides a general framework for linking genetic variants to gene dysregulation in disease.
Copy rights belong to original authors. Visit the link for more info
Podcast created by Paper Player, LLC