If you're processing NGS data, this is the Python wrapper around htslib you'll reach for constantly. It handles SAM/BAM/CRAM alignments, VCF/BCF variants, and FASTA/FASTQ sequences with a clean interface for fetching regions, running pileups, and calculating coverage. The 0-based coordinates take a minute to internalize (except when using region strings, which are 1-based like samtools), but once you do, it's straightforward. You'll need index files for random access, which is standard for this domain. It really shines when you're chaining operations across file types, like validating variants against aligned reads or extracting reference context around mutation sites.
npx skills add https://github.com/k-dense-ai/scientific-agent-skills --skill pysam