Identification of genetic variants that impact gene co-expression relationships using large-scale single-cell data

Abstract

BackgroundExpression quantitative trait loci (eQTL) studies have shown how genetic variants affect downstream gene expression. To identify the upstream regulatory processes, single-cell data can be used. Single-cell data also offers the unique opportunity to reconstruct personalized co-expression networks—by exploiting the large number of cells per individual, we can identify SNPs that alter co-expression patterns (co-expression QTLs, co-eQTLs) using a limited number of individuals.ResultsTo tackle the large multiple testing burden associated with a genome-wide analysis (i.e. the need to assess all combinations of SNPs and gene pairs), we conducted a co-eQTL meta-analysis across four scRNA-seq peripheral blood mononuclear cell datasets from three studies (reflecting 173 unique participants and 1 million cells) using a novel filtering strategy followed by a permutation-based approach. Before analysis, we evaluated the co-expression patterns to be used for co-eQTL identification using different external resources. The subsequent analysis identified a robust set of cell-type-specific co-eQTLs for 72 independent SNPs that affect 946 gene pairs, which we then replicated in a large bulk cohort. These co-eQTLs provide novel insights into how disease-associated variants alter regulatory networks. For instance, one co-eQTL SNP, rs1131017, that is associated with several autoimmune diseases affects the co-expression of RPS26 with other ribosomal genes. Interestingly, specifically in T cells, the SNP additionally affects co-expression of RPS26 and a group of genes associated with T cell-activation and autoimmune disease. Among these genes, we identified enrichment for targets of five T-cell-activation-related transcriptional factors whose binding sites harbor rs1131017. This reveals a previously overlooked process and pinpoints potential regulators that could explain the association of rs1131017 with autoimmune diseases.ConclusionOur co-eQTL results highlight the importance of studying gene regulation at the context-specific level to understand the biological implications of genetic variation. With the expected growth of sc-eQTL datasets, our strategy—combined with our technical guidelines—will soon identify many more co-eQTLs, further helping to elucidate unknown disease mechanisms.

Association of RPS26 gene polymorphism with different types of diabetes in Chinese individuals

“…In alignment with our earlier reference to the impact of RPS26 on T lymphocytes, it has been established that rs1131017 is a prominent SNP of a trans-regulatory site that affects multiple genes in CD4 + and CD8 + T lymphocytes 20 . In T lymphocytes, rs1131017 can influence several genes connected with T lymphocyte activation and autoimmune disease 24 , yielding a potential explanation for the linkage of this variant with type 1 diabetes mellitus and LADA.…”

“…Previous GWASs have associated these variants with vitiligo 21 , type 1 diabetes mellitus in white people 18 and rheumatoid arthritis 22 , respectively. Furthermore, rs1131017 has been identified as a lead SNP in a trans‐acting regulatory region within T lymphocytes 20 , affecting not only RPS26 expression 23 , but also the expression of genes associated with T lymphocyte activation and autoimmune disorders 24 . Although existing research suggests that the expression of the RPS26 gene might not be the molecular trait directly responsible for susceptibility to type 1 diabetes mellitus in white people 23 , the correlation between the rs1131017 polymorphism in the RPS26 gene and susceptibility to type 1 diabetes mellitus and other diabetes subtypes in the Chinese population remains unexplored, and warrants further investigation.…”

A spectral framework to map QTLs affecting joint differential networks of gene co-expression

“…Recent studies have extended the concept of eQTL to co-QTLs [30, 1, 40, 14]. These methods aim to identify genetic loci that explain coordinated changes in expression between pairs of genes.…”

“…One of the challenges is the massive number of statistical tests needed, which increases exponentially with the number of gene pairs analyzed. Some methods restrict co-QTL searches to previously identified eQTLs [30, 14], while others prioritize gene pairs based on prior knowledge [40]. These approaches reduce testing burdens but may miss important co-QTLs.…”

“…These approaches reduce testing burdens but may miss important co-QTLs. Furthermore, current co-QTL methods are limited by assuming linear models with additive effects [30, 1, 40, 14]. The additive assumption neglects dominance, recessiveness, or even transgressive inheritance, hindering the ability to capture the full genetic influence on co-expression networks.…”