Module parallel 

Parallel processing helpers (requires the parallel feature).

Two strategies are available:

• par_for_each and friends parse on one thread and hand batches of records to a rayon pool. This works with any reader, including gzip and standard input, and is the right default when the work per record is non-trivial.
• par_map_chunks_file splits an uncompressed file into byte ranges, snaps each range to a record boundary and parses the ranges in parallel. This is the only way to make parsing itself scale across cores.

use fastx::{FastxReader, parallel};

let data = b">a\nACGT\n>b\nGGCC\n>c\nAAAA\n";
let gc: Vec<Option<f64>> =
    parallel::par_map(&mut FastxReader::new(&data[..]), 64, |r| r.gc_content())?;
assert_eq!(gc, vec![Some(0.5), Some(1.0), Some(0.0)]);

Constants

DEFAULT_CHUNK_SIZE

Default number of records handed to the pool at a time.

Functions

par_fold

Fold every record into a single value in parallel.

par_for_each

Apply f to every record, in parallel across batches.

par_for_each_file

Apply f to every record of an uncompressed file with parallel parsing.

par_map

Map every record through f in parallel, preserving input order.

par_map_chunks_file

Run f on a reader for each chunk of an uncompressed file, in parallel.

par_stats

Compute SeqStats using all cores.

par_stats_file

Compute SeqStats for an uncompressed file with parallel parsing.

split_file

Split an uncompressed FASTA/FASTQ file into parts byte ranges that each start exactly on a record boundary.

Type Aliases

ChunkReader

Reader type handed to each worker by par_map_chunks_file.