denovoAssemblage/index.ts

/*
 * Require in path : zgrep, grep
 */

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

const async_exec = (prog: string, args: string[], onData: Function, onErr: 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 => onErr(data.toString().trim()))

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

const rmList = (list: string[]) => {
    return Promise.all(list.map(e => fs.promises.rm(e, {recursive: true})))
}

const isDone = (rsltDir:string) => {
    let done = false
    if(fs.existsSync(path.join(rsltDir, 'contigs.fasta'))) {
        if(fs.statSync(path.join(rsltDir, 'contigs.fasta')).size > 1) {
            done = true
        }
    }
    return done
}

const denovoAssemblage = (
    reads : string | string[],
    rnames: string[],
    spadesPath: string
) => {
    return new Promise<string>(async (resolve, reject) => {
        let readsIn: string[] = []
        let isPairedEnd = false
        const threads = String(os.cpus().length)

        if (Array.isArray(reads)) {
            if (reads.length > 2) {
                reject('Only R1 and R2 path are required')
            } else {
                console.log('Assuming paired end reads');
                isPairedEnd = true
                readsIn = reads
            }
        } else {
            readsIn = [reads]
        }

        // Prepare the fastq files for spades inputs
        const rnamesSel = '\'' + rnames.join('\\|') + '\''
        const tmpSubReads: string[] = []
        let isfq = []
    
        for (const R of readsIn) {
            const parsedPath = path.parse(R)
            const greper = parsedPath.ext === '.gz' ? 'zgrep' : 'grep'
            isfq.push(R.match(/\.fq/) ? true : false)
            const tmp = path.join(os.tmpdir(), (+new Date)  + '_' + parsedPath.name + (isfq ? '.fq' : '.fa'))
            tmpSubReads.push(tmp)
            await async_exec(greper, [rnamesSel, R, '-A3', '--no-group-separator', '>', tmp], console.log, console.log)
            
        }

        const correction = (isfq.filter(e=>e).length !== isfq.length || isfq.filter(e=>e).length === 0) ? '--only-assembler' : ''

        const args = isPairedEnd ? ['-1', tmpSubReads[0], '-2', tmpSubReads[1]] : ['-s', tmpSubReads[0]]
        const rsltDir = path.join(os.tmpdir(), (+new Date)  + '_spades')
        
        await async_exec(spadesPath, ['-t', threads, '--isolate', correction, ...args, '-o', rsltDir], console.log, console.log)

        if(!isDone(rsltDir)) {
            await rmList([rsltDir])
            console.log('Trying meta')
            await async_exec(spadesPath, ['-t', threads, '--meta', correction, ...args, '-o', rsltDir], console.log, console.log)
            if(!isDone(rsltDir)) {
                await rmList([rsltDir, ...tmpSubReads])
                reject('No convergence')
            }
        }

        await rmList(tmpSubReads)
        resolve(rsltDir)
    })
}

export { denovoAssemblage }

/*
(async()=>{
    const spadesPath = '/home/thomas/NGS/tools/SPAdes-3.15.0-Linux/bin/spades.py'
    const reads2 = '/home/thomas/Documents/Programmes/ttest/71-mer.fa' 
    const rnames = ['33530080_3',   '30971394_3',   '77190111_3',   '89825138_3',
      '22481866_3',   '111937620_4',  '24308941_6',   '27758147_3',
      '87242990_14',  '41688638_4',   '114699822_3',  '48573844_4',
      '91080996_3',   '99644261_3',   '77207124_3',   '28986564_10',
      '84400117_4',   '10884880_6',   '116082011_12', '1739367_3',
      '13550404_3',   '68446023_10',  '50560660_3',   '9046992_3']
    console.log(await denovoAssemblage(reads2, rnames, spadesPath));
})()
*/