Fast Pangenome Matching Statistics in Compressed Space with Move Structures

ms2 computes matching statistics and SMEMs in compressed pangenome indexes up to 48× faster than PHONI on 100 human chromosome 22 haplotypes, with similar index size.

• Why it matters: Exact matching statistics and SMEMs are crucial for homology search, but existing compressed pangenome methods are too slow compared with fast approximations like PMLs.
• What they did: ms2 uses move structures combined with LF, {Psi}, {varphi}, and {varphi}-1 permutations plus LCP summaries and samples to compute matching statistics and SMEMs efficiently in compressed space.
• The result: On HPRC2 (472 haplotypes) ms2 is at least 22× faster than ropebwt3, reaching 25–34× when reporting SMEM offsets, enabling real-time throughput for ~21 PromethION flow cells per thread.

AI summary of the abstract · may contain errors · not medical advice · read the paper.