DeCAF

DeCAF identifies cell-fraction quantitative trait loci (cfQTLs) in tumors by integrating allelic and total expression to resolve cell-type-specific genetic effects from bulk RNA-seq data.

Key Features:

• cfQTL detection: Identifies cell-fraction quantitative trait loci (cfQTLs) in tumors using both allelic and total expression information.
• Allelic and total expression integration: Integrates allelic expression data with overall gene expression levels to discern cell-type-specific genetic effects.
• Improved sensitivity and specificity: Demonstrates enhanced sensitivity and specificity compared to conventional interaction-eQTL mapping techniques.
• Bulk RNA-seq compatibility: Operates on bulk RNA-seq / RNA sequencing (RNA-seq) datasets, including applications to The Cancer Genome Atlas (TCGA) data.
• Discovery statistics: Identified 3,664 genes associated with cfQTLs across 14 distinct cell types at a false discovery rate (FDR) of 10%.
• Quantified improvement: Achieves a 5.63-fold improvement in cfQTL discovery over conventional methods.
• Cancer-risk enrichment: cfQTLs detected by DeCAF show greater enrichment for associations with cancer risk compared to conventional eQTLs.
• Tumor deconvolution: Leverages allelic and total expression to deconvolute complex tumor environments into specific cellular contributions.

Scientific Applications:

• Cell-type-specific QTL mapping in tumors: Discovery of cfQTLs that reveal genetic effects specific to particular cell types within heterogeneous tumor samples.
• Analysis of TCGA RNA-seq data: Application to The Cancer Genome Atlas RNA-seq datasets to identify cfQTLs across tumor cohorts.
• Prioritization of cancer-associated variants: Enrichment of cfQTLs for cancer risk associations enables prioritization of biologically meaningful variants.
• Studies of moderately sized cohorts: Enables exploration of cell-type-specific genetic effects in moderately sized studies using bulk RNA-seq data.

Methodology:

Integration of allelic expression data with overall gene expression levels applied to RNA sequencing (RNA-seq) datasets from The Cancer Genome Atlas (TCGA), with identification of 3,664 genes across 14 cell types at FDR 10% and comparison to interaction-eQTL mapping showing a 5.63-fold improvement.