Varlap: Bringing Alignment-Based Variant Quality Control to Scale
Year: 2026
Type: article
Abstract: Large-scale sequencing studies, particularly in cancer genomics, generate thousands of candidate variants, many arising from problematic alignment contexts such as low depth of coverage, sample contamination, or proximity to other variants. Manually inspecting these in tools such as IGV is effective for small variant sets but impractical at cohort scale. Varlap (github.com/bjpop/varlap) is an open-source tool that automates this process, computing per-variant alignment-context metrics from one or more BAM files, including edit distance, base and mapping quality, alignment length, clipping, and strand distribution. It has supported variant QC across thousands of samples in the Pan Prostate Cancer Group (panprostate.org) and in published studies of breast tumour mutational signatures. Since it makes no assumptions about the origin of the variants or reads, it is also generalizable beyond human and cancer genomics to other organisms and variant-calling contexts. Varlap's original Python implementation queries reads independently for each variant position. This performs well for sparsely distributed somatic variants but scales poorly when variants are closely spaced or when scanning contiguous genomic windows, since overlapping reads are repeatedly refetched from the BAM file. To address this bottleneck, Varlap has been reimplemented in Rust (github.com/jchchiu/rust-varlap), restructuring the core algorithm to iterate over reads rather than variant positions, so that each overlapping read is fetched at most once per genomic bin. This sets the foundation for substantially higher performance on dense or contiguous variant sets, while preserving Varlap's original output format and metrics, and lays the groundwork for future multithreaded execution.
Institutions: The University of Melbourne, University of Copenhagen
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7233608206
FWCI:
Cites:
Cited by:
Related to:
