ReQTL: identifying correlations between expressed SNVs and gene expression using RNA-sequencing data

Abstract

Motivation By testing for associations between DNA genotypes and gene expression levels, expression quantitative trait locus (eQTL) analyses have been instrumental in understanding how thousands of single nucleotide variants (SNVs) may affect gene expression. As compared to DNA genotypes, RNA genetic variation represents a phenotypic trait that reflects the actual allele content of the studied system. RNA genetic variation at expressed SNV loci can be estimated using the proportion of alleles bearing the variant nucleotide (variant allele fraction, VAFRNA). VAFRNA is a continuous measure which allows for precise allele quantitation in loci where the RNA alleles do not scale with the genotype count. We describe a method to correlate VAFRNA with gene expression and assess its ability to identify genetically regulated expression solely from RNA-sequencing (RNA-seq) datasets. Results We introduce ReQTL, an eQTL modification which substitutes the DNA allele count for the variant allele fraction at expressed SNV loci in the transcriptome (VAFRNA). We exemplify the method on sets of RNA-seq data from human tissues obtained though the Genotype-Tissue Expression (GTEx) project and demonstrate that ReQTL analyses are computationally feasible and can identify a subset of expressed eQTL loci. Availability and implementation A toolkit to perform ReQTL analyses is available at https://github.com/HorvathLab/ReQTL. Supplementary information Supplementary data are available at Bioinformatics online.

Estimating allele-specific expression of SNVs from 10x Genomics Single-Cell RNA-Sequencing Data

“…In scRNA studies, where the different cells are commonly in gradual states of progressive processes, VAFRNA analyses can be adopted to study lineages and cellular dynamics. Third, VAFRNA can be used to study functional SNVs from sets where matched DNA (and, respectively genotypes) is not available [24,25]. Ultimately, these analyses apply to expressed SNVs and will not capture loci positioned in transcriptionally silent regions.…”

“…These include loci exhibiting preferential expression of functional alleles, somatic mutations in cancer, and RNA-editing loci. Second, in contrast to the (static) genotypes, VAFRNA is dynamic and reflects the actual allele content in the system at any particular time, which allows for the assessment of dynamic and progressive processes [23][24][25]. Importantly, through primarily reflecting genetic variation, VAFRNA is an essential component of the genomic interactome and plays a major role in phenotype formation [26][27][28][29][30].…”

“…We have selected this platform due to its growing popularity along with: (1) high throughput (our analysis includes 26,640 cells obtained from three healthy donors), (2) high depth of sequencing (~150,000 sequencing reads per cell), and, (3) support for unique molecular identifiers (UMI) for removal of PCR-related sequencing bias. Because VAFRNA is sensitive to allele-mapping bias [24,25], we use SNV-aware alignments where reads mapped ambiguously due to the variant nucleotide are removed [32]. From the SNV-aware alignments, we systematically assess the ability to estimate VAFRNA using three different thresholds for minimum required number of unique sequencing reads (minR) -3, 5 and 10.…”

RsQTL: correlation of expressed SNVs with splicing using RNA-sequencing data

“…We have previously presented a systematic analysis of VAF RNA usage in QTL pipelines (Spurr, L. et al , 2019). Briefly, there are several important considerations for RsQTL applications.…”

“…eQTLs and sQTLs are both traditionally assessed from matched DNA and RNA datasets, where DNA is used for genotype estimation and RNA for expression (eQTLs), or splicing (sQTLs) estimation. We have recently developed a method to assess eQTLs from RNA-seq data alone-ReQTL (Spurr, L. et al , 2019) - which replaces the genotypes with the variant allele fraction, VAFrna, and identifies a subset of eQTLs. We present a related method, RsQTL ( R NA- sQTL ), which identifies splicing QTLs via correlation of VAFrna with the proportion of excised introns (percent spliced in, PSI) at loci with differential intron excision (Li, Y. I. et al , 2018).…”