eQTLseq

eQTLseq identifies genetic variants that influence gene expression by mapping expression quantitative trait loci (eQTLs) from paired DNA-seq and RNA-seq next-generation sequencing (NGS) data within populations.

Key Features:

• NGS paired-data support: Analyzes paired DNA-seq and RNA-seq assays from next-generation sequencing to detect multiple eQTLs simultaneously within a population.
• Hierarchical Probabilistic Models: Employs hierarchical probabilistic models to analyse multiple gene/variant associations from NGS data.
• Model Types: Supports transformation-based models that rely on transformations of RNA-seq data and count-based models that represent digital gene expression explicitly.
• Tractable Count-Based Modeling: Introduces latent variables within a sparse Bayesian modeling framework to improve estimation tractability for count-based models.
• Preprocessing Transformations: Incorporates arcsin, logit, and Laplace smoothing transformations for RNA-seq read counts as preprocessing steps.
• Benchmarking: Benchmarked on natural and simulated data from the 1000 Genomes and gEUVADIS projects across noise and model-assumption violation scenarios.
• Model Comparisons: Shows that an arcsin transformation of Laplace-smoothed data performs comparably to state-of-the-art models in small sample sizes and that an over-dispersed Poisson model is as effective as the Negative Binomial while offering easier estimation.

Scientific Applications:

• eQTL mapping for complex traits: Maps expression quantitative trait loci to elucidate the genetic architecture of complex traits.
• Analyses across data characteristics and sample sizes: Applicable to studies requiring transformation-based or count-based modeling across different experimental designs, including small sample sizes.

Methodology:

Utilizes hierarchical probabilistic models to analyse gene/variant associations; implements latent variables in a sparse Bayesian framework for tractable count-based estimation; and applies transformations such as arcsin, logit, and Laplace smoothing to RNA-seq read counts.