| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203 |
- /*
- * 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,
- log: Function
- ) => {
- 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: boolean[] = []
-
- 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[isfq.length-1] ? '.fq' : '.fa'))
- tmpSubReads.push(tmp)
- await async_exec(greper, [rnamesSel, R, '-A3', '--no-group-separator', '>', tmp], log, 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], log, log)
- if(!isDone(rsltDir) && isPairedEnd) {
- await rmList([rsltDir])
- console.log('Trying meta')
- await async_exec(spadesPath, ['-t', threads, '--meta', correction, ...args, '-o', rsltDir], log, log)
-
- if(!isDone(rsltDir)) {
- await rmList([rsltDir, ...tmpSubReads])
- reject('No convergence')
- } else {
- await rmList(tmpSubReads)
- resolve(rsltDir)
- }
- } else if (!isDone(rsltDir) && !isPairedEnd) {
- await rmList([rsltDir, ...tmpSubReads])
- reject('No convergence')
- } else if (isDone(rsltDir)) {
- 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/R1r.fq', '/home/thomas/Documents/Programmes/ttest/R2r.fq']
- const rnames = [
- 'A00680:166:HYCYGDMXX:1:1212:20528:10520_1',
- 'A00680:166:HYCYGDMXX:2:2184:24442:19680_2',
- 'A00680:166:HYCYGDMXX:1:1419:9344:19586_1',
- 'A00680:166:HYCYGDMXX:1:1306:14479:14387_1',
- 'A00680:166:HYCYGDMXX:2:2184:24442:19680_1',
- 'A00680:166:HYCYGDMXX:2:2148:4779:14653',
- 'A00680:166:HYCYGDMXX:2:2148:4860:14826',
- 'A00680:166:HYCYGDMXX:2:2148:4924:14935',
- 'A00680:166:HYCYGDMXX:2:2148:4933:14920',
- 'A00680:166:HYCYGDMXX:2:1205:29559:23249_2',
- 'A00680:166:HYCYGDMXX:2:1205:29559:23249',
- 'A00680:166:HYCYGDMXX:1:1209:1497:28354',
- 'A00680:166:HYCYGDMXX:2:2177:24171:24189_1',
- 'A00680:166:HYCYGDMXX:2:2221:30635:15875',
- 'A00680:166:HYCYGDMXX:2:1336:8034:35916_2',
- 'A00680:166:HYCYGDMXX:2:1336:8052:35916_2',
- 'A00680:166:HYCYGDMXX:2:1336:8034:35916',
- 'A00680:166:HYCYGDMXX:2:1336:8052:35916',
- 'A00680:166:HYCYGDMXX:2:1355:24758:35759',
- 'A00680:166:HYCYGDMXX:2:1356:27398:2879',
- 'A00680:166:HYCYGDMXX:2:1360:31955:31187_2',
- 'A00680:166:HYCYGDMXX:1:2342:28999:6715_2',
- 'A00680:166:HYCYGDMXX:1:2342:28999:6715',
- 'A00680:166:HYCYGDMXX:1:2213:1072:3818_2',
- 'A00680:166:HYCYGDMXX:1:2213:1072:3818',
- 'A00680:166:HYCYGDMXX:1:2310:10764:5572_2',
- 'A00680:166:HYCYGDMXX:1:2442:17671:31407_2',
- 'A00680:166:HYCYGDMXX:2:1409:10122:17613_2',
- 'A00680:166:HYCYGDMXX:2:1409:10981:19977_2',
- 'A00680:166:HYCYGDMXX:2:2414:23764:19038_2',
- 'A00680:166:HYCYGDMXX:1:1262:3323:12164',
- 'A00680:166:HYCYGDMXX:2:2414:23764:19038',
- 'A00680:166:HYCYGDMXX:1:1467:31105:14215',
- 'A00680:166:HYCYGDMXX:2:2212:10926:6417',
- 'A00680:166:HYCYGDMXX:1:1262:3323:12164_1',
- 'A00680:166:HYCYGDMXX:1:1157:13214:20932_2',
- 'A00680:166:HYCYGDMXX:1:1157:13223:20948_2',
- 'A00680:166:HYCYGDMXX:1:1157:14145:20040_2',
- 'A00680:166:HYCYGDMXX:2:1360:2356:14465',
- 'A00680:166:HYCYGDMXX:2:1360:2356:14465_1',
- 'A00680:166:HYCYGDMXX:1:1220:28492:17237',
- 'A00680:166:HYCYGDMXX:1:2220:28782:13573',
- 'A00680:166:HYCYGDMXX:2:1448:26024:25958',
- 'A00680:166:HYCYGDMXX:2:2420:7175:19774',
- 'A00680:166:HYCYGDMXX:2:1317:17372:23437',
- 'A00680:166:HYCYGDMXX:2:2420:10673:32002_2',
- 'A00680:166:HYCYGDMXX:2:2420:10673:32002',
- 'A00680:166:HYCYGDMXX:2:2420:10999:32847_2',
- 'A00680:166:HYCYGDMXX:2:2420:10999:32847',
- 'A00680:166:HYCYGDMXX:2:2420:9697:32628_2',
- 'A00680:166:HYCYGDMXX:2:2420:9697:32628',
- 'A00680:166:HYCYGDMXX:2:2229:25012:10582_2',
- 'A00680:166:HYCYGDMXX:2:2229:25012:10582',
- 'A00680:166:HYCYGDMXX:2:2229:25165:9846_2',
- 'A00680:166:HYCYGDMXX:2:2229:25165:9846',
- 'A00680:166:HYCYGDMXX:2:2229:25328:10128_2',
- 'A00680:166:HYCYGDMXX:2:2229:25328:10128',
- 'A00680:166:HYCYGDMXX:2:2229:25165:9846_1',
- 'A00680:166:HYCYGDMXX:2:1317:17372:23437_1',
- 'A00680:166:HYCYGDMXX:1:2176:11071:20697_2',
- 'A00680:166:HYCYGDMXX:1:2176:11071:20697',
- 'A00680:166:HYCYGDMXX:2:2132:9534:32064',
- 'A00680:166:HYCYGDMXX:1:2105:13187:20165_2',
- 'A00680:166:HYCYGDMXX:1:2105:13187:20165',
- 'A00680:166:HYCYGDMXX:1:2105:15365:20713',
- 'A00680:166:HYCYGDMXX:1:2105:16559:20462',
- 'A00680:166:HYCYGDMXX:1:2463:16152:4883_2',
- 'A00680:166:HYCYGDMXX:1:2463:18656:8844_2',
- 'A00680:166:HYCYGDMXX:2:2318:21694:26224',
- 'A00680:166:HYCYGDMXX:2:2318:22209:27993',
- 'A00680:166:HYCYGDMXX:2:1319:15908:27320_2',
- 'A00680:166:HYCYGDMXX:1:1462:5665:12774_1',
- 'A00680:166:HYCYGDMXX:1:2144:14796:8547_1',
- 'A00680:166:HYCYGDMXX:1:2144:14796:8547',
- 'A00680:166:HYCYGDMXX:1:1149:20139:3270',
- 'A00680:166:HYCYGDMXX:1:1149:24343:11490',
- 'A00680:166:HYCYGDMXX:1:1458:12228:29434_2',
- 'A00680:166:HYCYGDMXX:1:1458:12228:29434',
- 'A00680:166:HYCYGDMXX:1:1462:5665:12774',
- 'A00680:166:HYCYGDMXX:1:2105:15365:20713_2',
- 'A00680:166:HYCYGDMXX:1:2105:16559:20462_2',
- 'A00680:166:HYCYGDMXX:1:2144:14796:8547_2',
- 'A00680:166:HYCYGDMXX:2:1367:20166:33974_2',
- 'A00680:166:HYCYGDMXX:2:2301:7066:16141',
- 'A00680:166:HYCYGDMXX:2:2258:4227:16892',
- 'A00680:166:HYCYGDMXX:2:2229:25012:10582_1',
- 'A00680:166:HYCYGDMXX:2:1228:29939:33176',
- 'A00680:166:HYCYGDMXX:1:1425:24975:12587',
- 'A00680:166:HYCYGDMXX:1:2450:13874:8265',
- 'A00680:166:HYCYGDMXX:2:1367:20166:33974'
- ]
- const uniq_reads = [...new Set(rnames.map(e => e.split(/_[1-2]$/)[0]))];
- console.log(await denovoAssemblage(reads2, uniq_reads, spadesPath, console.log));
- })()*/
|