Task-Geometry Alignment: a design principle forparameter-free, accurate genomic search

Abstract

Standard alignment-based DNA barcoding falters because rigid geometric assumptions clash with biological realities, particularly insertions/deletions (indels) and fragment-to-reference asymmetry. To address this, we propose Task-Geometry Alignment (TGA), a design principle that structurally aligns the algorithm with the intrinsic geometry of the biological task. We implement TGA in TGAO_SCPLOWLIGNC_SCPLOW, a parameter-free tool that tiles reference databases into windows, applies gap-robust syncmer sketching, and indexes them for approximate nearest neighbor search. Across five benchmarks encompassing COI, 16S, and ITS markers, TGAO_SCPLOWLIGNC_SCPLOW achieves 9% higher accuracy and up to 10x faster query speeds than the leading aligner. Crucially, ablation studies confirm this architectural tiling is the causal driver of performance, allowing the tool to automatically match expert-parameterized configurations on fragment identification tasks. AvailabilitySource code and data are available at https://github.com/JustinBooneLab/TGAlign