@datagrok/bio - npm Package Compare versions

Comparing version 2.11.37 to 2.11.38

.eslintrc.json

@@ -18,3 +18,3 @@ {
   "rules": {
-    "indent": ["error", 2, {"SwitchCase": 1}],
+    "indent": ["error", 2, {"SwitchCase": 0}],
     "max-len": ["error", 120],
@@ -29,4 +29,3 @@ "no-unused-vars": "off",
       "error",
-      "multi-or-nest",
-      "consistent"
+      "multi-or-nest"
     ],
@@ -33,0 +32,0 @@ "brace-style": [

CHANGELOG.md

 # Bio changelog
 
+## 2.11.38 (2024-03-08)
+
+### Bug fixes
+
+* GROK-14910: PepSeA verbose output
+* MSA ensures docker container for PepSeA
+* Sample files harmonized
+
 ## 2.11.37 (2024-03-08)
@@ -4,0 +12,0 @@

package.json

@@ -8,3 +8,3 @@ {
   },
-  "version": "2.11.37",
+  "version": "2.11.38",
   "description": "Bioinformatics support (import/export of sequences, conversion, visualization, analysis). [See more](https://github.com/datagrok-ai/public/blob/master/packages/Bio/README.md) for details.",
@@ -38,3 +38,3 @@ "repository": {
     "@biowasm/aioli": "^3.1.0",
-    "@datagrok-libraries/bio": "5.39.26",
+    "@datagrok-libraries/bio": "5.39.29",
     "@datagrok-libraries/chem-meta": "^1.2.1",
@@ -41,0 +41,0 @@ "@datagrok-libraries/ml": "^6.4.10",

src/demo/bio01-similarity-diversity.ts

@@ -11,3 +11,3 @@ import * as grok from 'datagrok-api/grok';
 
-const dataFn: string = 'data/sample_FASTA_PT_activity.csv';
+const dataFn: string = 'samples/FASTA_PT_activity.csv';
 
@@ -14,0 +14,0 @@ export async function demoBio01UI() {

src/demo/bio01a-hierarchical-clustering-and-sequence-space.ts

@@ -12,3 +12,3 @@ import * as grok from 'datagrok-api/grok';
 
-const dataFn = 'data/sample_FASTA_PT_activity.csv';
+const dataFn = 'samples/FASTA_PT_activity.csv';
 const seqColName = 'sequence';
@@ -15,0 +15,0 @@

src/demo/bio01b-hierarchical-clustering-and-activity-cliffs.ts

@@ -17,3 +17,3 @@ import * as grok from 'datagrok-api/grok';
 
-const dataFn: string = 'data/sample_FASTA_PT_activity.csv';
+const dataFn: string = 'samples/FASTA_PT_activity.csv';
 
@@ -20,0 +20,0 @@ export async function demoBio01bUI() {

src/package.ts

@@ -915,3 +915,3 @@ /* Do not change these import lines to match external modules in webpack configuration */
     const app = new WebLogoApp(urlParams, 'webLogoAggApp');
-    const df: DG.DataFrame = await _package.files.readCsv('data/sample_FASTA_PT_activity.csv');
+    const df: DG.DataFrame = await _package.files.readCsv('samples/FASTA_PT_activity.csv');
     await grok.data.detectSemanticTypes(df);
@@ -941,3 +941,3 @@ await app.init(df);
     const urlParams = new URLSearchParams(window.location.search);
-    const df = await _package.files.readCsv('data/sample_HELM_empty_vals.csv');
+    const df = await _package.files.readCsv('samples/HELM_empty_vals.csv');
     const app = new GetRegionApp(urlParams, 'getRegionHelmApp');
@@ -944,0 +944,0 @@ await app.init({df: df, colName: 'HELM'});

src/tests/activity-cliffs-tests.ts

@@ -75,3 +75,3 @@ import * as grok from 'datagrok-api/grok';
   test('Helm', async () => {
-    const df = await _package.files.readCsv('data/sample_HELM_50.csv');
+    const df = await _package.files.readCsv('samples/HELM_50.csv');
     const _view = grok.shell.addTableView(df);
@@ -78,0 +78,0 @@

src/tests/detectors-tests.ts

@@ -158,4 +158,4 @@ import * as grok from 'datagrok-api/grok';
   const samples: { [key: string]: string } = {
-    [Samples.fastaFasta]: 'System:AppData/Bio/data/sample_FASTA.fasta',
-    [Samples.fastaPtCsv]: 'System:AppData/Bio/data/sample_FASTA_PT.csv',
+    [Samples.fastaFasta]: 'System:AppData/Bio/samples/FASTA.fasta',
+    [Samples.fastaPtCsv]: 'System:AppData/Bio/samples/FASTA_PT.csv',
     [Samples.msaComplex]: 'System:AppData/Bio/samples/MSA.csv',
@@ -162,0 +162,0 @@ [Samples.fastaCsv]: 'System:AppData/Bio/samples/FASTA.csv',

src/tests/msa-tests.ts

@@ -9,3 +9,3 @@ import * as grok from 'datagrok-api/grok';
 import {multipleSequenceAlignmentUI} from '../utils/multiple-sequence-alignment-ui';
-import {awaitContainerStart} from './utils';
+import {awaitContainerStart} from '../utils/docker';
 
@@ -12,0 +12,0 @@ category('MSA', async () => {

src/tests/pepsea-tests.ts

 import * as DG from 'datagrok-api/dg';
 
-import {before, category, expect, test} from '@datagrok-libraries/utils/src/test';
+import {before, category, expect, expectArray, test} from '@datagrok-libraries/utils/src/test';
 import {runPepsea} from '../utils/pepsea';
-import {awaitContainerStart} from './utils';
+import {TestLogger} from './utils/test-logger';
+import {errInfo} from '@datagrok-libraries/bio/src/utils/err-info';
+import {awaitContainerStart} from '../utils/docker';
 
 category('PepSeA', () => {
   const testCsv = `HELM,MSA
-  "PEPTIDE1{F.L.R.G.W.[MeF].Y.S.N.N.C}$$$$","F.L.R.G.W.MeF.Y..S.N.N.C"
-  "PEPTIDE1{F.L.R.G.Y.[MeF].Y.W.N.C}$$$$","F.L.R.G.Y.MeF.Y.W...N.C"
-  "PEPTIDE1{F.G.Y.[MeF].Y.W.S.D.N.C}$$$$","F...G.Y.MeF.Y.W.S.D.N.C"
-  "PEPTIDE1{F.L.R.G.Y.[MeF].Y.W.S.N.D.C}$$$$","F.L.R.G.Y.MeF.Y.W.S.N.D.C"
-  "PEPTIDE1{F.V.R.G.Y.[MeF].Y.W.S.N.C}$$$$","F.V.R.G.Y.MeF.Y.W.S..N.C"`;
+"PEPTIDE1{F.L.R.G.W.[MeF].Y.S.N.N.C}$$$$","F.L.R.G.W.MeF.Y..S.N.N.C"
+"PEPTIDE1{F.L.R.G.Y.[MeF].Y.W.N.C}$$$$","F.L.R.G.Y.MeF.Y.W...N.C"
+"PEPTIDE1{F.G.Y.[MeF].Y.W.S.D.N.C}$$$$","F...G.Y.MeF.Y.W.S.D.N.C"
+"PEPTIDE1{F.L.R.G.Y.[MeF].Y.W.S.N.D.C}$$$$","F.L.R.G.Y.MeF.Y.W.S.N.D.C"
+"PEPTIDE1{F.V.R.G.Y.[MeF].Y.W.S.N.C}$$$$","F.V.R.G.Y.MeF.Y.W.S..N.C"
+`;
 
+  const pepseaStderrCsv: string = `HELM,MSA
+"PEPTIDE1{F.L.Mis.G.W.[MeF].Y.S.N.N.C}$$$$","F.L.Mis.G.W.MeF.Y..S.N.N.C"
+"PEPTIDE1{F.L.Mis.G.Y.[MeF].Y.W.N.C}$$$$","F.L.Mis.G.Y.MeF.Y...W.N.C"
+"PEPTIDE1{F.G.Y.[MeF].Y.W.S.D.N.C}$$$$","F...G.Y.MeF.Y.W.S.D.N.C"
+`;
+  const pepseaStderrWarningList: string = 'Mis not found in Monomer Map\nMeF not found in Monomer Map\n';
+
+  const pepseaErrorCsv: string = `HELM
+"PEPTIDE1{[NH2].*.A.Q.T.T.Y.K.N.Y.R.R.N.L.L.*.[COOH]}$$$$"
+"PEPTIDE1{[NH2].M.A.N.T.T.Y.K.N.Y.R.N.N.L.L.*.[COOH]}$$$$"
+"PEPTIDE1{[NH2].*.A.N.T.T.Y.K.C.Y.R.R.N.L.L.*.[COOH]}$$$$"
+"PEPTIDE1{[NH2].*.A.N.T.T.Y.K.F.Y.R.R.N.L.L.*.[COOH]}$$$$"
+`;
+  const pepseaErrorError: string = 'PepSeA error: The pair (*,M) couldn\'t be found in the substitution matrix';
+
   before(async () => {
@@ -20,9 +38,31 @@ await awaitContainerStart();
   test('Basic alignment', async () => {
-    const table = DG.DataFrame.fromCsv(testCsv);
-    const alignedCol = await runPepsea(table.getCol('HELM'), 'msa(HELM)');
-    expect(alignedCol !== null, true, 'PepSeA container has not started');
-    const alignedTestCol = table.getCol('MSA');
-    for (let i = 0; i < alignedCol!.length; ++i)
-      expect(alignedCol!.get(i) == alignedTestCol.get(i), true);
+    const df = DG.DataFrame.fromCsv(testCsv);
+    const resMsaCol = await runPepsea(df.getCol('HELM'), 'msa(HELM)');
+    const tgtMsaCol = df.getCol('MSA');
+    for (let i = 0; i < resMsaCol!.length; ++i)
+      expect(resMsaCol!.get(i) == tgtMsaCol.get(i), true);
   }, {timeout: 60000 /* docker */});
+
+  test('stderr', async () => {
+    const logger = new TestLogger();
+    const df = DG.DataFrame.fromCsv(pepseaStderrCsv);
+    const resMsaCol = await runPepsea(df.getCol('HELM'), 'msa(HELM)',
+      undefined, undefined, undefined, undefined, logger);
+    const tgtMsaCol = df.getCol('MSA');
+    expectArray(resMsaCol!.toList(), tgtMsaCol.toList());
+    expect(logger.warningList[0].message, pepseaStderrWarningList);
+  }, {timeout: 60000 /* docker */});
+
+  test('error', async () => {
+    const logger = new TestLogger();
+    try {
+      const df = DG.DataFrame.fromCsv(pepseaErrorCsv);
+      const _resMsaCol = await runPepsea(df.getCol('HELM'), 'msa(HELM)',
+        undefined, undefined, undefined, undefined, logger);
+    } catch (err: any) {
+      const [errMsg, errStack] = errInfo(err);
+      logger.error(errMsg, undefined, errStack);
+    }
+    expect(logger.errorList[0].message, pepseaErrorError);
+  });
 });

src/tests/renderers-test.ts

@@ -74,3 +74,3 @@ import * as grok from 'datagrok-api/grok';
   async function _rendererMacromoleculeSeparator() {
-    const csv: string = await grok.dapi.files.readAsText('System:AppData/Bio/data/sample_SEPARATOR_PT.csv');
+    const csv: string = await grok.dapi.files.readAsText('System:AppData/Bio/samples/SEPARATOR_PT.csv');
     const df: DG.DataFrame = DG.DataFrame.fromCsv(csv);
@@ -117,3 +117,3 @@
   async function _testAfterMsa() {
-    const fastaTxt: string = await grok.dapi.files.readAsText('System:AppData/Bio/data/sample_FASTA.fasta');
+    const fastaTxt: string = await grok.dapi.files.readAsText('System:AppData/Bio/samples/FASTA.fasta');
     const df: DG.DataFrame = importFasta(fastaTxt)[0];
@@ -158,3 +158,3 @@
   async function _testAfterConvert() {
-    const csv: string = await grok.dapi.files.readAsText('System:AppData/Bio/data/sample_FASTA_PT.csv');
+    const csv: string = await grok.dapi.files.readAsText('System:AppData/Bio/samples/FASTA_PT.csv');
     const df: DG.DataFrame = DG.DataFrame.fromCsv(csv);
@@ -161,0 +161,0 @@

src/tests/utils.ts

@@ -37,10 +37,2 @@ import * as DG from 'datagrok-api/dg';
 
-/** Waits if container is not started
- * @param {number} ms - time to wait in milliseconds */
-export async function awaitContainerStart(ms: number = 30000): Promise<void> {
-  const dc = await grok.dapi.docker.dockerContainers.filter('bio').first();
-  const dcId = dc.id;
-  await startDockerContainer(dcId, ms);
-}
-
 export async function awaitGrid(grid: DG.Grid, timeout: number = 5000): Promise<void> {
@@ -47,0 +39,0 @@ await delay(0);

src/tests/viewers.ts

@@ -13,3 +13,3 @@ import * as DG from 'datagrok-api/dg';
     test(v, async () => {
-      const df = await readDataframe('data/sample_FASTA_DNA.csv');
+      const df = await readDataframe('samples/FASTA_DNA.csv');
       await testViewer(v, df, {detectSemanticTypes: true});
@@ -16,0 +16,0 @@ }, {

src/utils/docker.ts

 import * as grok from 'datagrok-api/grok';
 import * as ui from 'datagrok-api/ui';
 import * as DG from 'datagrok-api/dg';
+
 import {delay} from '@datagrok-libraries/utils/src/test';
 
-export async function startDockerContainer(dcId: string, timeout: number = 30000): Promise<void> {
-  // TODO: Use the new dockerContainers API
-  const res = await grok.dapi.docker.dockerContainers.run(dcId /*, true */);
+import {Pepsea} from './pepsea';
+
+/** Waits if container is not started
+ * @param {number} ms - time to wait in milliseconds */
+export async function awaitContainerStart(ms: number = 60000): Promise<void> {
+  const dc = await Pepsea.getDockerContainer();
+  await startDockerContainer(dc, ms);
+}
+
+export async function startDockerContainer(argDc: DG.DockerContainer, timeout: number = 60000): Promise<void> {
+  // argDc contains current container status
+  let dc: DG.DockerContainer | null = argDc;
   let end: boolean = false;
   for (let i = 0; i < timeout / 200; ++i) {
-    const dc = await grok.dapi.docker.dockerContainers.find(dcId);
+    if (dc === null) dc = await grok.dapi.docker.dockerContainers.find(argDc.id);
+
+    if (isStarted(dc)) {
+      end = true;
+      break;
+    }
+
     switch (dc.status) {
-      case 'stopped': {
-        await grok.dapi.docker.dockerContainers.run(dcId);
-        break;
-      }
-      case 'pending change':
-      case 'changing': {
-        // skip to wait
-        break;
-      }
-      case 'checking':
-      case 'started': {
-        end = true;
-        break;
-      }
-      case 'error': {
-        throw new Error('Docker container error state.');
-      }
+    case 'stopped': {
+      // TODO: Use the new dockerContainers API
+      await grok.dapi.docker.dockerContainers.run(dc.id);
+      break;
     }
-    if (end) break;
+    case 'pending change':
+    case 'changing': {
+      // skip to wait
+      break;
+    }
+    case 'error': {
+      throw new Error('Docker container error state.');
+    }
+    }
+    dc = null;
     await delay(200);
   }
-  if (!end) throw new Error('Docker container run timeout.');
+  if (!end) throw new Error('Docker container start timeout.');
   // this.dc = await grok.dapi.docker.dockerContainers.find(dcId);
 }
+
+export function isStarted(dc: DG.DockerContainer): boolean {
+  return dc.status === 'checking' || dc.status === 'started';
+}

src/utils/multiple-sequence-alignment-ui.ts

@@ -14,2 +14,3 @@ import * as grok from 'datagrok-api/grok';
 import {kalignVersion, msaDefaultOptions} from './constants';
+import {awaitContainerStart} from './docker';
 
@@ -72,3 +73,3 @@ import {_package} from '../package';
     //TODO: remove when the new version of datagrok-api is available
-    //TODO: allow only macromolecule colums to be chosen
+    //TODO: allow only macromolecule columns to be chosen
     const colInput = ui.columnInput('Sequence', table, seqCol, async () => {
@@ -171,4 +172,7 @@ performAlignment = await onColInputChange(
 
-      return async () => await runPepsea(col, unusedName, methodInput.value!,
-        gapOpenInput.value!, gapExtendInput.value!, clustersColInput.value);
+      return async () => {
+        await awaitContainerStart();
+        return runPepsea(col, unusedName, methodInput.value!,
+          gapOpenInput.value!, gapExtendInput.value!, clustersColInput.value);
+      };
     } else if (checkInputColumnUI(col, col.name, [NOTATION.SEPARATOR], [ALPHABET.UN], false)) {
@@ -183,4 +187,7 @@ //if the column is separator with unknown alphabet, it might be helm. check if it can be converted to helm
 
-      return async () => await runPepsea(helmCol, unusedName, methodInput.value!,
-        gapOpenInput.value!, gapExtendInput.value!, clustersColInput.value);
+      return async () => {
+        await awaitContainerStart();
+        return runPepsea(helmCol, unusedName, methodInput.value!,
+          gapOpenInput.value!, gapExtendInput.value!, clustersColInput.value);
+      };
     } else {
@@ -187,0 +194,0 @@ gapOpenInput.value = null;

src/utils/pepsea.ts

@@ -9,2 +9,3 @@ /* Do not change these import lines to match external modules in webpack configuration */
 import {NOTATION, TAGS as bioTAGS, ALIGNMENT, ALPHABET} from '@datagrok-libraries/bio/src/utils/macromolecule';
+import {ILogger} from '@datagrok-libraries/bio/src/utils/logger';
 
@@ -42,14 +43,12 @@ import * as C from './constants';
  * @param {DG.Column} clustersCol - The column containing the clusters of the sequences.
+ * @param logger {ILogger} Logger
  */
 export async function runPepsea(srcCol: DG.Column<string>, unUsedName: string,
   method: typeof pepseaMethods[number] = 'ginsi', gapOpen: number = 1.53, gapExtend: number = 0.0,
-  clustersCol: DG.Column<string | number> | null = null,
-): Promise<DG.Column<string> | null> {
+  clustersCol: DG.Column<string | number> | null = null, logger?: ILogger
+): Promise<DG.Column<string>> {
   const pepseaContainer = await Pepsea.getDockerContainer();
-  if (pepseaContainer.status !== 'started' && pepseaContainer.status !== 'checking') {
-    grok.log.warning('PepSeA container has not started yet');
-    return null;
-  }
+  if (pepseaContainer.status !== 'started' && pepseaContainer.status !== 'checking')
+    throw new Error('PepSeA container has not started yet');
 
-
   const peptideCount = srcCol.length;
@@ -77,3 +76,3 @@ clustersCol ??= DG.Column.int('Clusters', peptideCount).init(0);
   for (const body of bodies) { // getting aligned sequences for each cluster
-    const alignedObject = await requestAlignedObjects(pepseaContainer.id, body, method, gapOpen, gapExtend);
+    const alignedObject = await requestAlignedObjects(pepseaContainer.id, body, method, gapOpen, gapExtend, logger);
     const alignments = alignedObject.Alignment;
@@ -100,4 +99,5 @@
 
-async function requestAlignedObjects(dockerfileId: string, body: PepseaBodyUnit[], method: string, gapOpen: number,
-  gapExtend: number): Promise<PepseaResponse> {
+async function requestAlignedObjects(
+  dockerfileId: string, body: PepseaBodyUnit[], method: string, gapOpen: number, gapExtend: number, logger?: ILogger
+): Promise<PepseaResponse> {
   const params = {
@@ -109,4 +109,51 @@ method: 'POST',
   const path = `/align?method=${method}&gap_open=${gapOpen}&gap_extend=${gapExtend}`;
-  const response = await grok.dapi.docker.dockerContainers.request(dockerfileId, path, params);
-  return JSON.parse(response ?? '{}');
+  let responseObj: any;
+  if ('fetchProxy' in grok.dapi.docker.dockerContainers) {
+    // new dockerContainers API
+    const t1: number = window.performance.now();
+    // @ts-ignore
+    const response: Response = await grok.dapi.docker.dockerContainers.fetchProxy(dockerfileId, path, params);
+    const t2: number = window.performance.now();
+    _package.logger.debug(`Bio: requestAlignedObjects() dockerContainers.fetchProxy(), ET: ${(t2 - t1)} ms`);
+    const responseContentType = response.headers.get('content-type');
+    const isJson: boolean = responseContentType === 'application/json';
+    if (!response.ok && isJson) {
+      const responseJson = await response.json();
+      const pepseaErrorMsg = responseJson['pepsea-error'];
+      if (!!pepseaErrorMsg)
+        throw new Error(`PepSeA error: ${pepseaErrorMsg}`);
+
+      const datagrokErrorMsg = responseJson['datagrok-error'];
+      if (!!datagrokErrorMsg)
+        throw new Error(`Datagrok error: ${datagrokErrorMsg}`);
+
+      throw new Error(response.statusText);
+    } else if (!response.ok && !isJson) {
+      const responseStr = await response.text();
+      throw new Error(`Error: ${responseStr}`);
+    } else if (!isJson) {
+      const responseStr = await response.text();
+      throw new Error(`Error: PepSeA expected JSON response, got '${responseStr}'.`);
+    }
+    responseObj = await response.json();
+  } else {
+    const responseStr = await grok.dapi.docker.dockerContainers.request(dockerfileId, path, params)!;
+    if (!responseStr)
+      throw new Error('Empty response');
+    responseObj = JSON.parse(responseStr);
+
+    const pepseaErrorMsg = responseObj['pepsea-error'];
+    if (!!pepseaErrorMsg)
+      throw new Error(`PepSeA error: ${pepseaErrorMsg}`);
+
+    const datagrokErrorMsg = responseObj['datagrok-error'];
+    if (!!datagrokErrorMsg)
+      throw new Error(`Datagrok error: ${datagrokErrorMsg}`);
+  }
+  // Check for pepsea stderr output
+  if ('pepsea-stderr' in responseObj) {
+    const pepseaStdErr: string = responseObj['pepsea-stderr'] as string;
+    logger?.warning(pepseaStdErr);
+  }
+  return responseObj as PepseaResponse;
 }

No alert changes

Worsened metrics

Total package byte prevSize

20686549

decreased by-1.05%

Number of package files

228

decreased by-0.87%

Lines of code

38748

decreased by-1.66%

Dependency changes

• + Added@datagrok-libraries/bio@5.39.29(transitive)

• - Removed@datagrok-libraries/bio@5.39.26(transitive)

• Updated@datagrok-libraries/bio@5.39.29