Home

lra - a long read alignment platform. This project is in early phases. For robust, well-supported alignment, consider https://github.com/lh3/minimap2.

Much of the purpose of lsa is to test different approaches to long read alignment, while maintaining the speed of current methods. lsa uses a two-tiered seeding and global-chaining approach, where a rough alignment is determined from large exact matches, and then refined using smaller (7-9 bp) matches.

The global seeds are currently found using minimizers. However, if all k-mers in a read are matched against all (k,w) minimizers in a genome, the total runtime is ~15-20% slower, indicating that other sub-sampling techniques may be used to index a genome with marginal drop in performance.