Integrating single-cell sQTL mapping with deep-learning splicing prediction identifies causal variants under...

Area of Science:

• Human genetics
• Molecular biology
• Computational biology

Background:

• Identifying causal variants is crucial for understanding human genetics and disease.
• Molecular quantitative trait locus (molQTL) analyses, including expression QTL (eQTL) and splicing QTL (sQTL), link genetic variants to molecular phenotypes.
• Pinpointing causal variants and their regulatory effects in molQTL studies remains a challenge.

Purpose of the Study:

• To integrate splicing QTL (sQTL) analysis with deep learning-based splicing effect annotation.
• To identify causal genetic variants and elucidate their functional effects on human phenotypes.
• To provide mechanistic insights from genotype to disease susceptibility.

Main Methods:

• Applied a single-cell GWAS method (scHi-HOST) across lymphoblastoid cell lines with and without influenza A virus (IAV) infection.
• Integrated sQTL mapping with AI-based splice prediction and statistical fine-mapping.
• Experimentally validated a causal variant (rs2297616) in PARP2 using CRISPR-mediated editing.

Main Results:

• Identified ~43,000 sQTL SNP-junction pairs associated with 217 genes during IAV infection.
• Prioritized 57 likely causal variants affecting cis-acting splicing components.
• Validated rs2297616's effect on PARP2 protein isoforms and IAV levels; linked 57 sQTLs to over 100 GWAS traits, including autoimmune diseases.

Conclusions:

• Generated a catalog of causal sQTL with direct splicing impacts.
• Provided causal mechanistic insights connecting genotype to disease susceptibility.
• Highlighted the role of splicing variants in complex human diseases like autoimmune conditions.