SegmentQTL: Identifying genetic variants influencing molecular phenotypes in copy number-driven cancers

Motivation: Molecular quantitative trait loci (molQTL) analysis links genetic variants to molecular phenotypes, such as gene expression, but existing tools do not account for the structural complexity of copy number-driven cancers. High genomic instability of these cancers leads to chromosomal breaks (breakpoints), which disrupt the physical connection between genes and adjacent regulatory elements. Standard molQTL methods are unable to accommodate breakpoint information and would therefore indiscriminately test associations across breakpoints, leading to spurious signals and reduced biological relevance. To address these challenges, we developed SegmentQTL, a segmentation-aware molQTL analysis tool, designed to improve the accuracy of association testing in unstable cancer genomes by incorporating sample-specific breakpoint information. Results: SegmentQTL applies an integrated purifying filtering step that removes associations spanning breakpoints, ensuring that only variants within the same segment as the phenotype are tested. This prevents false discoveries and reduces background noise. We evaluated SegmentQTL on selected genes from stable and unstable genomic regions and compared its results with a previously published state-of-the-art tool. In stable regions, SegmentQTL produced similar results, validating its approach. In unstable regions, however, the filtering step refined detected associations by shifting peak locations and removing artefactual signals that would arise if genomic instability were not properly accounted for. Availability and implementation: https://github.com/HautaniemiLab/SegmentQTL