Mapping Splicing Quantitative Trait Loci in RNA-Seq

Abstract

BACKGROUNDOne of the major mechanisms of generating mRNA diversity is alternative splicing, a regulated process that allows for the flexibility of producing functionally different proteins from the same genomic sequences. This process is often altered in cancer cells to produce aberrant proteins that drive the progression of cancer. A better understanding of the misregulation of alternative splicing will shed light on the development of novel targets for pharmacological interventions of cancer.METHODSIn this study, we evaluated three statistical methods, random effects meta-regression, beta regression, and generalized linear mixed effects model, for the analysis of splicing quantitative trait loci (sQTL) using RNA-Seq data. All the three methods use exon-inclusion levels estimated by the PennSeq algorithm, a statistical method that utilizes paired-end reads and accounts for non-uniform sequencing coverage.RESULTSUsing both simulated and real RNA-Seq datasets, we compared these three methods with GLiMMPS, a recently developed method for sQTL analysis. Our results indicate that the most reliable and powerful method was the random effects meta-regression approach, which identified sQTLs at low false discovery rates but higher power when compared to GLiMMPS.CONCLUSIONSWe have evaluated three statistical methods for the analysis of sQTLs in RNA-Seq. Results from our study will be instructive for researchers in selecting the appropriate statistical methods for sQTL analysis.

The Expanding Landscape of Alternative Splicing Variation in Human Populations

“…Several computational methods have been developed for identifying sQTLs from population-scale genotype and RNA-seq data. 57 , 58 , 59 , 60 , 61 Zhao et al. developed GLiMMPS, a computational method that identifies sQTLs at the event level by associating the PSI values of individual alternative splicing events with genotypes across the population.…”

DRIMSeq: a Dirichlet-multinomial framework for multivariate count outcomes in genomics

“…Such events capture not only cassette exons but also alternative 3’ and 5’ splice sites, mutually exclusive exons or intron retention. GLiMMPS 32 and Jia et al 33 , with quantification from PennSeq 34 , use event inclusion levels for detecting SNPs that are associated with differential splicing. However, there are (hypothetical) instances where changes in splicing pattern may not be captured by exon-level quantifications (Figure 1A in the paper by Monlog et al 35 ).…”

“…Similarly, separate modeling and testing of exon junctions ( Altrans 27 ) or splicing events ( rMATS 29 , GLiMMPS 32 , Jia et al 33 , Montgomery et al 49 ) of a gene leads to non-independent statistical tests, although the full effect of this on calibration (e.g., controlling the rate of false discoveries) is not known. Nevertheless, with the larger number of tests, the multiple testing correction becomes more extreme.…”

CyTOF workflow: differential discovery in high-throughput high-dimensional cytometry datasets

“…Nevertheless, as in genomic data analysis, pure logistic regression is not able to capture the overdispersion that is present in HDCyto data. A natural extension to model the extra variation is the generalized linear mixed model (GLMM), where the random effect is defined by the sample ID (observation-level random effects 44,45 ). Additionally, in our example the patient pairing could be accounted in the model by incorporating a random intercept for each patient.…”