index.js 9.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200
  1. "use strict";
  2. /*
  3. * Require in path : zgrep, grep
  4. */
  5. var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, generator) {
  6. function adopt(value) { return value instanceof P ? value : new P(function (resolve) { resolve(value); }); }
  7. return new (P || (P = Promise))(function (resolve, reject) {
  8. function fulfilled(value) { try { step(generator.next(value)); } catch (e) { reject(e); } }
  9. function rejected(value) { try { step(generator["throw"](value)); } catch (e) { reject(e); } }
  10. function step(result) { result.done ? resolve(result.value) : adopt(result.value).then(fulfilled, rejected); }
  11. step((generator = generator.apply(thisArg, _arguments || [])).next());
  12. });
  13. };
  14. var __importDefault = (this && this.__importDefault) || function (mod) {
  15. return (mod && mod.__esModule) ? mod : { "default": mod };
  16. };
  17. Object.defineProperty(exports, "__esModule", { value: true });
  18. exports.denovoAssemblage = void 0;
  19. const child_process_1 = require("child_process");
  20. const os_1 = __importDefault(require("os"));
  21. const path_1 = __importDefault(require("path"));
  22. const fs_1 = __importDefault(require("fs"));
  23. const async_exec = (prog, args, onData, onErr) => {
  24. return new Promise((resolve, reject) => {
  25. const child = (0, child_process_1.spawn)(prog, args, { shell: true });
  26. child.stdout.on('data', data => onData(data.toString().trim()));
  27. child.stderr.on('data', data => onErr(data.toString().trim()));
  28. child.on('error', err => reject(err));
  29. child.on('exit', code => resolve(code));
  30. });
  31. };
  32. const rmList = (list) => {
  33. return Promise.all(list.map(e => fs_1.default.promises.rm(e, { recursive: true })));
  34. };
  35. const isDone = (rsltDir) => {
  36. let done = false;
  37. if (fs_1.default.existsSync(path_1.default.join(rsltDir, 'contigs.fasta'))) {
  38. if (fs_1.default.statSync(path_1.default.join(rsltDir, 'contigs.fasta')).size > 1) {
  39. done = true;
  40. }
  41. }
  42. return done;
  43. };
  44. const denovoAssemblage = (reads, rnames, spadesPath, log) => {
  45. return new Promise((resolve, reject) => __awaiter(void 0, void 0, void 0, function* () {
  46. let readsIn = [];
  47. let isPairedEnd = false;
  48. const threads = String(os_1.default.cpus().length);
  49. if (Array.isArray(reads)) {
  50. if (reads.length > 2) {
  51. reject('Only R1 and R2 path are required');
  52. }
  53. else {
  54. console.log('Assuming paired end reads');
  55. isPairedEnd = true;
  56. readsIn = reads;
  57. }
  58. }
  59. else {
  60. readsIn = [reads];
  61. }
  62. // Prepare the fastq files for spades inputs
  63. const rnamesSel = '\'' + rnames.join('\\|') + '\'';
  64. const tmpSubReads = [];
  65. let isfq = [];
  66. for (const R of readsIn) {
  67. const parsedPath = path_1.default.parse(R);
  68. const greper = parsedPath.ext === '.gz' ? 'zgrep' : 'grep';
  69. isfq.push(R.match(/\.fq/) ? true : false);
  70. const tmp = path_1.default.join(os_1.default.tmpdir(), (+new Date) + '_' + parsedPath.name + (isfq[isfq.length - 1] ? '.fq' : '.fa'));
  71. tmpSubReads.push(tmp);
  72. yield async_exec(greper, [rnamesSel, R, '-A3', '--no-group-separator', '>', tmp], log, log);
  73. }
  74. const correction = (isfq.filter(e => e).length !== isfq.length || isfq.filter(e => e).length === 0) ? '--only-assembler' : '';
  75. const args = isPairedEnd ? ['-1', tmpSubReads[0], '-2', tmpSubReads[1]] : ['-s', tmpSubReads[0]];
  76. const rsltDir = path_1.default.join(os_1.default.tmpdir(), (+new Date) + '_spades');
  77. yield async_exec(spadesPath, ['-t', threads, '--isolate', correction, ...args, '-o', rsltDir], log, log);
  78. if (!isDone(rsltDir) && isPairedEnd) {
  79. yield rmList([rsltDir]);
  80. console.log('Trying meta');
  81. yield async_exec(spadesPath, ['-t', threads, '--meta', correction, ...args, '-o', rsltDir], log, log);
  82. if (!isDone(rsltDir)) {
  83. yield rmList([rsltDir, ...tmpSubReads]);
  84. reject('No convergence');
  85. }
  86. else {
  87. yield rmList(tmpSubReads);
  88. resolve(rsltDir);
  89. }
  90. }
  91. else if (!isDone(rsltDir) && !isPairedEnd) {
  92. yield rmList([rsltDir, ...tmpSubReads]);
  93. reject('No convergence');
  94. }
  95. else if (isDone(rsltDir)) {
  96. yield rmList(tmpSubReads);
  97. resolve(rsltDir);
  98. }
  99. }));
  100. };
  101. exports.denovoAssemblage = denovoAssemblage;
  102. /*
  103. (async()=>{
  104. const spadesPath = '/home/thomas/NGS/tools/SPAdes-3.15.0-Linux/bin/spades.py'
  105. const reads2 = ['/home/thomas/Documents/Programmes/ttest/R1r.fq', '/home/thomas/Documents/Programmes/ttest/R2r.fq']
  106. const rnames = [
  107. 'A00680:166:HYCYGDMXX:1:1212:20528:10520_1',
  108. 'A00680:166:HYCYGDMXX:2:2184:24442:19680_2',
  109. 'A00680:166:HYCYGDMXX:1:1419:9344:19586_1',
  110. 'A00680:166:HYCYGDMXX:1:1306:14479:14387_1',
  111. 'A00680:166:HYCYGDMXX:2:2184:24442:19680_1',
  112. 'A00680:166:HYCYGDMXX:2:2148:4779:14653',
  113. 'A00680:166:HYCYGDMXX:2:2148:4860:14826',
  114. 'A00680:166:HYCYGDMXX:2:2148:4924:14935',
  115. 'A00680:166:HYCYGDMXX:2:2148:4933:14920',
  116. 'A00680:166:HYCYGDMXX:2:1205:29559:23249_2',
  117. 'A00680:166:HYCYGDMXX:2:1205:29559:23249',
  118. 'A00680:166:HYCYGDMXX:1:1209:1497:28354',
  119. 'A00680:166:HYCYGDMXX:2:2177:24171:24189_1',
  120. 'A00680:166:HYCYGDMXX:2:2221:30635:15875',
  121. 'A00680:166:HYCYGDMXX:2:1336:8034:35916_2',
  122. 'A00680:166:HYCYGDMXX:2:1336:8052:35916_2',
  123. 'A00680:166:HYCYGDMXX:2:1336:8034:35916',
  124. 'A00680:166:HYCYGDMXX:2:1336:8052:35916',
  125. 'A00680:166:HYCYGDMXX:2:1355:24758:35759',
  126. 'A00680:166:HYCYGDMXX:2:1356:27398:2879',
  127. 'A00680:166:HYCYGDMXX:2:1360:31955:31187_2',
  128. 'A00680:166:HYCYGDMXX:1:2342:28999:6715_2',
  129. 'A00680:166:HYCYGDMXX:1:2342:28999:6715',
  130. 'A00680:166:HYCYGDMXX:1:2213:1072:3818_2',
  131. 'A00680:166:HYCYGDMXX:1:2213:1072:3818',
  132. 'A00680:166:HYCYGDMXX:1:2310:10764:5572_2',
  133. 'A00680:166:HYCYGDMXX:1:2442:17671:31407_2',
  134. 'A00680:166:HYCYGDMXX:2:1409:10122:17613_2',
  135. 'A00680:166:HYCYGDMXX:2:1409:10981:19977_2',
  136. 'A00680:166:HYCYGDMXX:2:2414:23764:19038_2',
  137. 'A00680:166:HYCYGDMXX:1:1262:3323:12164',
  138. 'A00680:166:HYCYGDMXX:2:2414:23764:19038',
  139. 'A00680:166:HYCYGDMXX:1:1467:31105:14215',
  140. 'A00680:166:HYCYGDMXX:2:2212:10926:6417',
  141. 'A00680:166:HYCYGDMXX:1:1262:3323:12164_1',
  142. 'A00680:166:HYCYGDMXX:1:1157:13214:20932_2',
  143. 'A00680:166:HYCYGDMXX:1:1157:13223:20948_2',
  144. 'A00680:166:HYCYGDMXX:1:1157:14145:20040_2',
  145. 'A00680:166:HYCYGDMXX:2:1360:2356:14465',
  146. 'A00680:166:HYCYGDMXX:2:1360:2356:14465_1',
  147. 'A00680:166:HYCYGDMXX:1:1220:28492:17237',
  148. 'A00680:166:HYCYGDMXX:1:2220:28782:13573',
  149. 'A00680:166:HYCYGDMXX:2:1448:26024:25958',
  150. 'A00680:166:HYCYGDMXX:2:2420:7175:19774',
  151. 'A00680:166:HYCYGDMXX:2:1317:17372:23437',
  152. 'A00680:166:HYCYGDMXX:2:2420:10673:32002_2',
  153. 'A00680:166:HYCYGDMXX:2:2420:10673:32002',
  154. 'A00680:166:HYCYGDMXX:2:2420:10999:32847_2',
  155. 'A00680:166:HYCYGDMXX:2:2420:10999:32847',
  156. 'A00680:166:HYCYGDMXX:2:2420:9697:32628_2',
  157. 'A00680:166:HYCYGDMXX:2:2420:9697:32628',
  158. 'A00680:166:HYCYGDMXX:2:2229:25012:10582_2',
  159. 'A00680:166:HYCYGDMXX:2:2229:25012:10582',
  160. 'A00680:166:HYCYGDMXX:2:2229:25165:9846_2',
  161. 'A00680:166:HYCYGDMXX:2:2229:25165:9846',
  162. 'A00680:166:HYCYGDMXX:2:2229:25328:10128_2',
  163. 'A00680:166:HYCYGDMXX:2:2229:25328:10128',
  164. 'A00680:166:HYCYGDMXX:2:2229:25165:9846_1',
  165. 'A00680:166:HYCYGDMXX:2:1317:17372:23437_1',
  166. 'A00680:166:HYCYGDMXX:1:2176:11071:20697_2',
  167. 'A00680:166:HYCYGDMXX:1:2176:11071:20697',
  168. 'A00680:166:HYCYGDMXX:2:2132:9534:32064',
  169. 'A00680:166:HYCYGDMXX:1:2105:13187:20165_2',
  170. 'A00680:166:HYCYGDMXX:1:2105:13187:20165',
  171. 'A00680:166:HYCYGDMXX:1:2105:15365:20713',
  172. 'A00680:166:HYCYGDMXX:1:2105:16559:20462',
  173. 'A00680:166:HYCYGDMXX:1:2463:16152:4883_2',
  174. 'A00680:166:HYCYGDMXX:1:2463:18656:8844_2',
  175. 'A00680:166:HYCYGDMXX:2:2318:21694:26224',
  176. 'A00680:166:HYCYGDMXX:2:2318:22209:27993',
  177. 'A00680:166:HYCYGDMXX:2:1319:15908:27320_2',
  178. 'A00680:166:HYCYGDMXX:1:1462:5665:12774_1',
  179. 'A00680:166:HYCYGDMXX:1:2144:14796:8547_1',
  180. 'A00680:166:HYCYGDMXX:1:2144:14796:8547',
  181. 'A00680:166:HYCYGDMXX:1:1149:20139:3270',
  182. 'A00680:166:HYCYGDMXX:1:1149:24343:11490',
  183. 'A00680:166:HYCYGDMXX:1:1458:12228:29434_2',
  184. 'A00680:166:HYCYGDMXX:1:1458:12228:29434',
  185. 'A00680:166:HYCYGDMXX:1:1462:5665:12774',
  186. 'A00680:166:HYCYGDMXX:1:2105:15365:20713_2',
  187. 'A00680:166:HYCYGDMXX:1:2105:16559:20462_2',
  188. 'A00680:166:HYCYGDMXX:1:2144:14796:8547_2',
  189. 'A00680:166:HYCYGDMXX:2:1367:20166:33974_2',
  190. 'A00680:166:HYCYGDMXX:2:2301:7066:16141',
  191. 'A00680:166:HYCYGDMXX:2:2258:4227:16892',
  192. 'A00680:166:HYCYGDMXX:2:2229:25012:10582_1',
  193. 'A00680:166:HYCYGDMXX:2:1228:29939:33176',
  194. 'A00680:166:HYCYGDMXX:1:1425:24975:12587',
  195. 'A00680:166:HYCYGDMXX:1:2450:13874:8265',
  196. 'A00680:166:HYCYGDMXX:2:1367:20166:33974'
  197. ]
  198. const uniq_reads = [...new Set(rnames.map(e => e.split(/_[1-2]$/)[0]))];
  199. console.log(await denovoAssemblage(reads2, uniq_reads, spadesPath, console.log));
  200. })()*/