synpact: accurate, memory-light PacBio HiFi read mapping via a hierarchy of locally-consistent syncmer blocks

PacBio HiFi long-read mapping using a hierarchy of seed sizes built by Locally Consistent Parsing over syncmers, with sliding-window voting

The authors present synpact, a PacBio HiFi long-read mapper using several seed sizes organized in a hierarchy, built by Locally Consistent Parsing over syncmers, where a read is mapped by querying different levels followed by sliding-window voting. By storing only the coarse upper levels rather than the full hierarchy, the index holds several times fewer entries while handling errors by falling back to finer levels. On simulated HiFi data, synpact matches or approaches minimap2 accuracy with higher precision in most cases, while using roughly 5-13x less peak memory (e.g., ~0.8 GB vs ~10.7 GB on human) and mapping faster on large or repetitive genomes. On real HiFi reads, synpact shows high concordance with minimap2. The tool is written in Rust and available open source.

Synthesis written by Geno'X. For the full original abstract, please refer to the source publication.

An interesting algorithmic contribution tackling the real bottleneck of accurate long-read mappers: memory consumption. The 5-13x memory gain while retaining near-minimap2 accuracy is concrete, and open-source Rust eases adoption. It remains a preprint whose strongest results rest on simulated data; concordance on real reads will need confirmation on clinical data before integration into a diagnostic pipeline.

Analysis by Dr Thibaut Benquey