bigwig.cr

A native Crystal implementation based on libBigWig.

The code was generated with LLM coding agent.

Install

Add to shard.yml:

dependencies:
  bigwig:
    github: kojix2/bigwig

Read

require "bigwig"

BigWig.open("signal.bw") do |bw|
  bw.each_interval("chr1", 1_000...2_000) do |interval|
    puts "#{interval.start}\t#{interval.stop}\t#{interval.value}"
  end

  values = bw.values("chr1", 1_000, 2_000)
  means = bw.stats("chr1", 0, 1_000_000, type: :mean, bins: 100)
end

Coordinates are zero-based and half-open. values returns one Float32 per base, with NaN for gaps. stats returns Array(Float64); it may use zoom data for approximate results. Pass exact: true for exact results.

Use each_entry for bigBed:

BigWig.open("annotations.bb") do |bb|
  bb.each_entry("chr1", 1_000, 2_000) do |entry|
    puts "#{entry.start}\t#{entry.stop}\t#{entry.rest}"
  end
end

Write

chromosomes = {"chr1" => 248_956_422_i64}

BigWig.create("output.bw", chromosomes) do |bw|
  bw.add("chr1", 100, 200, 1.5)
  bw.add("chr1", 200, 300, 2.0)
end

Add intervals in chromosome declaration order, sorted and non-overlapping within each chromosome. Existing output files are preserved by default. Pass options: BigWig::WriteOptions.new(overwrite: true) to replace one. bigBed writing is not supported.

Input and output

BigWig.open accepts a local path, an HTTP(S) URL, or seekable IO. BigWig.create accepts a local path or IO. Caller-provided IO stays open unless sync_close: true is passed.

BigWig.open("https://example.org/signal.bw") do |bw|
  puts bw.max("chr1", 0, 100_000)
end

HTTP(S) reading requires a server that supports Range requests; there is no full-download fallback. It uses Crystal's standard HTTP::Client, which does not use HTTP_PROXY or HTTPS_PROXY automatically.

Command-line examples

The files in examples/ are standalone CLI tools. Run any of them with --help for all options.

crystal run examples/query_intervals.cr -- --list-chromosomes signal.bw
crystal run examples/query_intervals.cr -- --chrom chr1 --start 1000 --end 2000 signal.bw
crystal run examples/dump_values.cr -- --chrom chr1 --start 1000 --end 2000 --omit-gaps signal.bw
crystal run examples/bedgraph_to_bigwig.cr -- chrom.sizes input.bedgraph output.bw

query_intervals prints overlapping bigWig intervals or bigBed entries as tab-separated rows. dump_values prints chromosome, zero-based position, and value; it includes NaN gaps unless --omit-gaps is set. The converter accepts - instead of a bedGraph path to read standard input. Its input must be sorted by chromosome declaration order and contain no overlapping intervals.

Development

gem install bio-bigwig
ruby scripts/check.rb

scripts/check.rb is also the CI entry point. It runs specs, checks fixture bytes, compiles examples, writes three test files with this library, and verifies them with bio-bigwig. The reference report is saved to build/reference-validation.json. Regenerate fixtures with crystal run scripts/fixtures/generate.cr -- --write.

License

MIT.