Genome-wide association studies (GWAS) have achieved great success in identifying genomic loci robustly associated with complex traits. However, given the non-coding nature of the findings, the causal genes that mediate the associations are difficult to establish. To tackle this challenge, many approaches and tools have been proposed that integrate transcriptome and GWAS studies. Colocalization approaches seek to quantify the probability of shared causal variants, whereas association approaches test the correlation between the genetic component of expression traits with the complex traits. All these methods rely on high-quality QTL mapping, fine-mapping, and prediction strategies. However, current large-scale RNA-seq studies do not take full advantage of total and allele-specific reads due to the computational burden of existing methods. Here, we provide a unified framework that combines total and allele-specific (at the gene level) read counts and is scalable to large studies with thousands of samples, such as GTEx. Using simulated and real data from GTEx, we demonstrate the improved performance of our framework. For about a third of the genes, which had large enough allele-specific counts, we show an average gain of 29% in effective sample size. We implemented a suite of tools, mixQTL, mixFine, and mixPred, which are made freely available. ### Competing Interest Statement F.A. is an inventor on a patent application related to TensorQTL; H.K.I. has received speaker honoraria from GSK and AbbVie.

Download data

• Downloaded 401 times
• Download rankings, all-time:
  • Site-wide: 43,063 out of 94,912
  • In genomics: 3,876 out of 5,955
• Year to date:
  • Site-wide: 7,624 out of 94,912
• Since beginning of last month:
  • Site-wide: 14,451 out of 94,912

Altmetric data

Downloads over time

Distribution of downloads per paper, site-wide