Year: 2024
Type: conference-paper
Abstract: Bowtie 2 is a popular short sequence aligner used by many bioinformatics groups, often as part of the Qiita microbial study management platform. One of the key computational steps during alignment is the extraction of seeds from the read sequences and aligning them with the help of the FM Index. This step was memory access-bound, with access latency as the primary limitation due to a pseudo-random-access pattern. In this paper we present the algorithmic changes used and the performance improvements obtained by re-factoring the code to overlap compute with memory pre-fetching.
Authors: Igor Sfiligoi, Daniel McDonald, Rob Knight
Institution: University of California San Diego
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W4402673836