aligner/index.ts

import { spawn } from 'child_process';
import { cpus } from 'os';
import fs from 'fs'
import path from 'path';

const async_exec = (prog: string, args: string[], onData: Function) => {
    return new Promise((resolve, reject) => {
        const child = spawn(prog, args, {shell: true})

        child.stdout.on('data', data => onData(data.toString().trim()))
        child.stderr.on('data', data => onData(data.toString().trim()))

        child.on('error', err => reject(err))
        child.on('exit', code => resolve(code))
    })
}

const asyncBwaMem = (
    refPath   : string,
    R1        : string | Array<string>,
    R2        : string | Array<string>,
    runName   : string,
    libName   : string,
    output_dir: string,
    onData    : Function,
    options?   : any,
    ) => {

    if (typeof options === 'undefined') options = {}

    const refName = path.parse(refPath).name
  
    const bwa        = 'bwa'
    const samblaster = 'samblaster'
    const samtools   = 'samtools'
    const sambamba   = 'sambamba'
  
    const R1_arr = Array.isArray(R1) ? R1.join(' ') : R1
    const R2_arr = Array.isArray(R2) ? R2.join(' ') : R2
  
    const R1_kitty = R1_arr.slice(-2) === 'gz' ? 'zcat' : 'cat'
    const R2_kitty = R2_arr.slice(-2) === 'gz' ? 'zcat' : 'cat'
  
    const R1_in = `'< ${R1_kitty} ${R1_arr}'`
    const R2_in = `'< ${R2_kitty} ${R2_arr}'`
  
    const discordantFile = path.join(output_dir, `bwa_mem_discordants_on_${refName}.sam`)
    const splitterFile   = path.join(output_dir, `bwa_mem_splitters_on_${refName}.sam`)
    const unmappedFile   = path.join(output_dir, `bwa_mem_unmapped_on_${refName}.fq`)
    const bam            = path.join(output_dir, `bwa_mem_properly_on_${refName}.bam`)
    const bamSorted      = path.join(output_dir, `bwa_mem_properly_on_${refName}.sorted.bam`)
  
    return new Promise<string[]>(async (resolve, reject) => {
      try {
        const code = await async_exec(
        bwa, ['mem',
          '-t', String(cpus().length),
          '-R', `"@RG\\tPL:Illumina\\tID:${+(new Date)}\\tSM:${runName}\\tLB:${libName}"`,
          refPath, R1_in, R2_in,
        '|',
        samblaster,
          '--addMateTags', // https://github.com/GregoryFaust/samblaster
          '-a', // Accept duplicate marks already in input file
          '-e', // Exclude reads marked as duplicates from discordant, splitter, and/or unmapped
          '-d', discordantFile,
          '-s', splitterFile,
          '-u', unmappedFile,
        '|',
        samtools,
          'view',
          '-b',
          '-',
        '>',
        bam
        ], (message: string) => onData('[BWA-MEM] ' + message))
        onData('[BWA-MEM][EXIT CODE] ' + code)
        
        const code_sort = await async_exec(
          sambamba, ['sort',
            '-t', String(require('os').cpus().length),
            bam
        ], (message: string) => onData('[SAMBAMBA-SORT] ' + message))
        onData('[SAMBAMBA-SORT][EXIT CODE] ' + code_sort)
  
        fs.unlinkSync(bam)
  
        resolve([bamSorted, discordantFile, splitterFile, unmappedFile])
      } catch (err) {
        reject(err)
      }
    })
}

export { asyncBwaMem }