bed-js

Parses BED lines into feature objects, using the file's autoSql schema when it has one. Includes a standalone autoSql parser and the UCSC bigBed schemas.

Install

npm install @gmod/bed

Usage

Feed the parser one line at a time:

import BED from '@gmod/bed'

const parser = new BED()
const features = text
  .split('\n')
  .filter(line => line.trim() && !/^(#|track|browser)/.test(line))
  .map(line => parser.parseLine(line))

parser.parseLine('chr1\t0\t100\tgene1\t50\t+')
// { chrom: 'chr1', chromStart: 0, chromEnd: 100, name: 'gene1',
//   score: 50, strand: 1 }

parseLine treats every line as a feature, so a header or comment line comes back as a feature with NaN coordinates. Filter those lines out first.

parseLine takes lines from any source, such as a .bed.gz queried with @gmod/tabix or a bigBed read with @gmod/bbi.

When a file's first line names its columns, pass that line to the constructor as columnNames instead of dropping it with the other # lines:

const [header, ...rest] = text.split('\n')
const parser = new BED({ columnNames: header.replace(/^#/, '').split('\t') })
rest.map(line => parser.parseLine(line))

Schemas

The constructor takes one of four ways to name and type the columns:

new BED() // standard BED columns
new BED({ autoSql }) // an autoSql string, e.g. from a bigBed header
new BED({ type: 'bigGenePred' }) // a builtin UCSC schema
new BED({ columnNames }) // names from a '#chrom start end ...' header

Documentation

• API — constructor, parseLine, exported types
• Schemas — the builtin list, autoSql, column-name headers
• Parsing behavior — strand, missing data, BED12+n, and what the parser guesses when there is no schema

Academic Use

This package was written with funding from the NHGRI as part of the JBrowse project. If you use it in an academic project that you publish, please cite the most recent JBrowse paper, which will be linked from jbrowse.org.

License

MIT © Colin Diesh