This is a forked version of minimap2 which has the capability of using open syncmers (Edgar 2021, see https://peerj.com/articles/10805/), instead of minimizers for selecting k-mers to be used for seeding.
It is known that open syncmers perform better than minimizers for k-mer matching tasks under mutation. We show in Shaw and Yu (2021) that open syncmers perform well against other k-mer selection methods in theory and that it improves sensitivity of mapping, especially when reads are relatively divergent from the reference.
git clone https://github.com/bluenote-1577/os-minimap2
cd os-minimap2 && make
## Using open syncmers with (s,t) = (11,3).
./minimap2 -a --syncs 11 --synct 3 test/MT-human.fa test/MT-orang.fa > test.sam
To use syncmers, specify parameters --syncs (s) --synct (t) where t,s correspond to parameters in the definition of open syncmers. We require k > s. If these parameters are not specified, minimap2 will use minimizers instead.
For detailed instructions on how to use minimap2, see https://github.com/lh3/minimap2.
Currently homo-polymer compressed runs are not supported. Indexing and mapping as separate steps is also not supported.
Li, H. (2018). Minimap2: pairwise alignment for nucleotide sequences. Bioinformatics, 34:3094-3100. doi:10.1093/bioinformatics/bty191
Shaw, Jim and Yu, Yun William. "Theory of local k-mer selection with applications to long-read alignment" Bioinformatics (2022).