This does not simplify BWA — the parameters that matter are exposed, and both the MEM and backtrack paths are here, so choosing between them is still your call. What it removes is the shell: an assistant that can already reason about which algorithm suits your read length can now run it against the files you name and hand back the alignment.
A server around the BWA aligner. It builds the index for a reference genome, aligns reads with the BWA-MEM algorithm, and runs the older backtrack path — finding coordinates first, then producing single-end or paired-end SAM output.
- An index built for a reference genome, with the indexing algorithm chosen to suit its size
- Reads aligned with BWA-MEM, single or paired end, with seed length, band width, thread count and read group all settable
- The backtrack path: coordinates found with mismatch and gap-open limits under your control
- Single-end and paired-end SAM produced from those coordinates
- Large references and read files handled, with the file-size ceiling and command timeout yours to set
BWA installed on the machine and Python for the server. Everything runs locally against files on disk — no account, no key, and no sequence data leaves the machine.
Build from source — clone the repository and build it, then point your client at the binary