Safe, idiomatic Rust bindings for minibwa — a fast short-read, adaptive, and Hi-C DNA aligner.

Workflow

1. Build an index once from a FASTA with [Index::build_from_fasta].
2. Load the index with [Index::load]; the loaded index is Send + Sync and can be shared across threads.
3. Create an [Aligner] from &Index and &Opts — cheap, no allocation.
4. Create one [ThreadBuf] per worker thread — holds the per-thread scratch arena; it is Send but not Sync.
5. Call [Aligner::map] or [Aligner::map_pair] to align reads and get owned [Hit] vectors.

Methylation support

Pass [Meth::C2T] or [Meth::G2A] to [Aligner::map] for single-read bisulfite alignment, or enable [Opts::set_methylation] together with [Aligner::map_pair] for paired-end bisulfite alignment.

Paired-end support

Use [Aligner::map_pair] for paired-end alignment. Insert-size parameters can be tuned with [Opts::set_pe_insert_size].

Parallelism

Thread-level parallelism is caller-owned: create one [ThreadBuf] per thread and call [Aligner::map] / [Aligner::map_pair] concurrently. The shared [Index] is read-only during mapping.

Example

use minibwa::{Index, Opts, Aligner, ThreadBuf, Meth};
Index::build_from_fasta("ref.fa", "ref", false, 4)?;
let idx = Index::load("ref", false)?;
let opts = Opts::new();
let aligner = Aligner::new(&idx, &opts);
let mut buf = ThreadBuf::new();
for hit in aligner.map(&mut buf, "read1", b"ACGT...", Meth::None)? {
    println!(
        "{} {}..{} {}",
        hit.contig.as_deref().unwrap_or("*"),
        hit.ref_start,
        hit.ref_end,
        hit.cigar_string(),
    );
}
# Ok::<(), minibwa::Error>(())