@datagrok/bio 2.0.16 → 2.0.17

package/package.json

@@ -5,7 +5,7 @@
     "name": "Leonid Stolbov",
     "email": "lstolbov@datagrok.ai"
   },
-  "version": "2.0.16",
+  "version": "2.0.17",
   "description": "Bio is a [package](https://datagrok.ai/help/develop/develop#packages) for the [Datagrok](https://datagrok.ai) platform",
   "repository": {
     "type": "git",
@@ -13,18 +13,18 @@
     "directory": "packages/Bio"
   },
   "dependencies": {
-    "@biowasm/aioli": ">=2.4.0",
-    "@datagrok-libraries/bio": "^4.4.3",
-    "@datagrok-libraries/chem-meta": "1.0.0",
+    "@biowasm/aioli": "^3.1.0",
+    "@datagrok-libraries/bio": "^4.4.7",
+    "@datagrok-libraries/chem-meta": "1.0.1",
     "@datagrok-libraries/ml": "^6.2.0",
-    "@datagrok-libraries/utils": "^1.6.2",
+    "@datagrok-libraries/utils": "^1.10.1",
     "@deck.gl/core": "^8.7.5",
     "@deck.gl/layers": "^8.7.5",
     "@luma.gl/constants": "^8.5.10",
     "@luma.gl/core": "^8.5.10",
-    "@phylocanvas/phylocanvas.gl": "^1.43.0",
+    "@phylocanvas/phylocanvas.gl": "^1.44.0",
     "cash-dom": "latest",
-    "datagrok-api": "^1.6.12",
+    "datagrok-api": "^1.7.0",
     "dayjs": "^1.11.4",
     "openchemlib": "6.0.1",
     "rxjs": "^6.5.5",

package/src/package-test.ts

@@ -12,6 +12,7 @@ import './tests/splitters-test';
 import './tests/renderers-test';
 import './tests/convert-test';
 import './tests/fasta-handler-test';
+import './tests/fasta-export-tests';
 import './tests/WebLogo-positions-test';
 import './tests/checkInputColumn-tests';
 import './tests/similarity-diversity-tests';

package/src/package.ts

@@ -34,7 +34,8 @@ import {splitAlignedSequences} from '@datagrok-libraries/bio/src/utils/splitter'
 import * as C from './utils/constants';
 import {SequenceSimilarityViewer} from './analysis/sequence-similarity-viewer';
 import {SequenceDiversityViewer} from './analysis/sequence-diversity-viewer';
-import { substructureSearchDialog } from './substructure-search/substructure-search';
+import {substructureSearchDialog} from './substructure-search/substructure-search';
+import {saveAsFastaUI} from './utils/save-as-fasta';
 
 //tags: init
 export async function initBio() {
@@ -516,3 +517,9 @@ export function bioSubstructureSearch(col: DG.Column): void {
   substructureSearchDialog(col);
 }
 
+//name: saveAsFasta
+//description: As FASTA...
+//tags: fileExporter
+export function saveAsFasta() {
+  saveAsFastaUI();
+}

package/src/tests/WebLogo-positions-test.ts

@@ -41,7 +41,7 @@ ATC-G-TTGC--
     seqCol.setTag(DG.TAGS.UNITS, NOTATION.FASTA);
     seqCol.setTag(UnitsHandler.TAGS.alphabet, ALPHABET.DNA);
 
-    const wlViewer: WebLogo = await df.plot.fromType('WebLogo') as unknown as WebLogo;
+    const wlViewer: WebLogo = (await df.plot.fromType('WebLogo')) as WebLogo;
     tv.dockManager.dock(wlViewer.root, DG.DOCK_TYPE.DOWN);
 
     tvList.push(tv);
@@ -94,8 +94,8 @@ ATC-G-TTGC--
       return i > 2;
     });
     df.filter.fireChanged();
-    const wlViewer: WebLogo = await df.plot.fromType('WebLogo', {'shrinkEmptyTail': true}) as unknown as WebLogo;
-
+    const wlViewer: WebLogo = (await df.plot.fromType('WebLogo',
+      {'shrinkEmptyTail': true})) as WebLogo;
     tv.dockManager.dock(wlViewer.root, DG.DOCK_TYPE.DOWN);
 
     tvList.push(tv);
@@ -134,7 +134,8 @@ ATC-G-TTGC--
     seqCol.setTag(DG.TAGS.UNITS, NOTATION.FASTA);
     seqCol.setTag(UnitsHandler.TAGS.alphabet, ALPHABET.DNA);
 
-    const wlViewer: WebLogo = await df.plot.fromType('WebLogo', {'skipEmptyPositions': true}) as unknown as WebLogo;
+    const wlViewer: WebLogo = (await df.plot.fromType('WebLogo',
+      {'skipEmptyPositions': true})) as WebLogo;
     tv.dockManager.dock(wlViewer.root, DG.DOCK_TYPE.DOWN);
 
     tvList.push(tv);

package/src/tests/convert-test.ts

@@ -1,7 +1,8 @@
-import {category, expectArray, test} from '@datagrok-libraries/utils/src/test';
-
-import * as grok from 'datagrok-api/grok';
 import * as DG from 'datagrok-api/dg';
+import * as ui from 'datagrok-api/ui';
+import * as grok from 'datagrok-api/grok';
+
+import {category, expect, expectArray, test} from '@datagrok-libraries/utils/src/test';
 
 import {ConverterFunc} from './types';
 import {NotationConverter} from '@datagrok-libraries/bio/src/utils/notation-converter';
@@ -139,6 +140,7 @@ RNA1{P.R(U)P.R(U)P.R(C)P.R(A)P.R(A)P.R(C)P.P.P}$$$
     return function(srcCol: DG.Column): DG.Column {
       const converter = new NotationConverter(srcCol);
       const resCol = converter.convert(tgtNotation, tgtSeparator);
+      expect(resCol.getTag('units'), tgtNotation);
       return resCol;
     };
   };

package/src/tests/fasta-export-tests.ts

@@ -0,0 +1,110 @@
+import * as DG from 'datagrok-api/dg';
+import * as ui from 'datagrok-api/ui';
+import * as grok from 'datagrok-api/grok';
+
+import {category, expect, expectArray, test} from '@datagrok-libraries/utils/src/test';
+import {saveAsFastaDo, wrapSequence} from '../utils/save-as-fasta';
+import {WebLogo} from '@datagrok-libraries/bio/src/viewers/web-logo';
+
+type SaveAsFastaTestArgs = { srcCsv: string, idCols: string [], seqCol: string, lineWidth: number, tgtFasta: string };
+
+category('fastaExport', () => {
+
+  enum WrapDataTest {
+    single = 'single',
+    multi = 'multi'
+  }
+
+  const wrapData: { [key: string]: { src: string, tgt: string[] } } = {
+    [WrapDataTest.single]: {
+      src: 'MDYKETLLMPKTDFPMRGGLP',
+      tgt: ['MDYKETLLMP', 'KTDFPMRGGL', 'P'],
+    },
+    [WrapDataTest.multi]: {
+      src: 'M[MeI]YKETLL[MeF]PKTDFPMRGGL[MeA]',
+      tgt: ['M[MeI]YKETLL[MeF]P', 'KTDFPMRGGL', '[MeA]'],
+    },
+  };
+
+  enum SaveAsFastaTests {
+    test1 = 'test1',
+    test2 = 'test2'
+  }
+
+  const saveAsFastaData: {
+    [key: string]: SaveAsFastaTestArgs
+  } = {
+    [SaveAsFastaTests.test1]: {
+      srcCsv: `id,seq
+1,MDYKETLLMP
+2,KTDFPMRGGL
+3,P`,
+      idCols: ['id'],
+      seqCol: 'seq',
+      lineWidth: 10,
+      tgtFasta: `>1
+MDYKETLLMP
+>2
+KTDFPMRGGL
+>3
+P
+`
+    },
+    [SaveAsFastaTests.test2]: {
+      srcCsv: `id,id2,seq
+1,seqA,M[MeI]YKETLL[MeF]P
+2,seqB,KTDFPMRGGL
+3,seqC,[MeA]
+`,
+      idCols: ['id2', 'id'],
+      seqCol: 'seq',
+      lineWidth: 5,
+      tgtFasta: `>seqA|1
+M[MeI]YKE
+TLL[MeF]P
+>seqB|2
+KTDFP
+MRGGL
+>seqC|3
+[MeA]
+`
+    }
+  };
+
+  test('wrapSequenceSingle', async () => {
+    _testWrapSequence(WrapDataTest.single, 10);
+  });
+
+  test('wrapSequenceMulti', async () => {
+    _testWrapSequence(WrapDataTest.multi, 10);
+  });
+
+  test('saveAsFastaTest1', async () => {
+    _testSaveAsFasta(saveAsFastaData[SaveAsFastaTests.test1]);
+  });
+
+  test('saveAsFastaTest2', async () => {
+    _testSaveAsFasta(saveAsFastaData[SaveAsFastaTests.test2]);
+  });
+
+  function _testWrapSequence(testKey: string, lineWidth: number = 10) {
+    const splitter = WebLogo.splitterAsFasta;
+
+    const srcSeq: string = wrapData[testKey].src;
+    const wrapRes: string[] = wrapSequence(srcSeq, splitter, lineWidth);
+    const wrapTgt: string[] = wrapData[testKey].tgt;
+
+    expectArray(wrapRes, wrapTgt);
+  }
+
+  async function _testSaveAsFasta(args: SaveAsFastaTestArgs) {
+    const df: DG.DataFrame = DG.DataFrame.fromCsv(args.srcCsv);
+
+    const seqCol: DG.Column = df.getCol(args.seqCol);
+    const idCols: DG.Column[] = args.idCols.map((colName) => df.getCol(colName));
+
+    const fastaRes: string = saveAsFastaDo(idCols, seqCol, args.lineWidth);
+    expect(fastaRes, args.tgtFasta);
+  }
+});
+

package/src/tests/splitters-test.ts

@@ -27,6 +27,11 @@ category('splitters', () => {
   const helm2 = 'PEPTIDE1{meI.hHis.Hcy.Q.T.W.Q.Phe_4NH2.D-Tyr_Et.Tyr_ab-dehydroMe.dV.E.N.N.meK}$$$';
 
   const data: { [key: string]: [string, string[]] } = {
+    fastaMulti: [
+      'M[MeI]YKETLL[MeF]PKTDFPMRGGL[MeA]',
+      ['M', 'MeI', 'Y', 'K', 'E', 'T', 'L', 'L', 'MeF', 'P',
+        'K', 'T', 'D', 'F', 'P', 'M', 'R', 'G', 'G', 'L', 'MeA']
+    ],
     helm1: [
       'PEPTIDE1{meI.hHis.Aca.N.T.dE.Thr_PO3H2.Aca.D-Tyr_Et.Tyr_ab-dehydroMe.dV.E.N.D-Orn.D-aThr.Phe_4Me}$$$',
       ['meI', 'hHis', 'Aca', 'N', 'T', 'dE', 'Thr_PO3H2', 'Aca', 'D-Tyr_Et',
@@ -59,6 +64,8 @@ category('splitters', () => {
     ],
   };
 
+  test('fastaMulti', async () => { await _testFastaSplitter(data.fastaMulti[0], data.fastaMulti[1]); });
+
   test('helm1', async () => { await _testHelmSplitter(data.helm1[0], data.helm1[1]); });
   test('helm2', async () => { await _testHelmSplitter(data.helm2[0], data.helm2[1]); });
   test('helm3-multichar', async () => { await _testHelmSplitter(data.helm3[0], data.helm3[1]); });
@@ -113,6 +120,12 @@ PEPTIDE1{hHis.Aca.Cys_SEt}$$$,5.72388
   });
 });
 
+export async function _testFastaSplitter(src: string, tgt: string[]) {
+  const res: string[] = WebLogo.splitterAsFasta(src);
+  console.debug(`Bio: tests: splitters: src=${JSON.stringify(src)}, res=${JSON.stringify(res)} .`);
+  expectArray(res, tgt);
+}
+
 export async function _testHelmSplitter(src: string, tgt: string[]) {
   const res: string[] = WebLogo.splitterAsHelm(src);
   console.debug(`Bio: tests: splitters: src=${JSON.stringify(src)}, res=${JSON.stringify(res)} .`);

package/src/utils/convert.ts

@@ -1,8 +1,8 @@
 import * as DG from 'datagrok-api/dg';
 import * as ui from 'datagrok-api/ui';
 import * as grok from 'datagrok-api/grok';
-import $ from 'cash-dom';
 
+import $ from 'cash-dom';
 import {Subscription} from 'rxjs';
 import {NotationConverter} from '@datagrok-libraries/bio/src/utils/notation-converter';
 import {NOTATION, UnitsHandler} from '@datagrok-libraries/bio/src/utils/units-handler';

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

@@ -36,11 +36,10 @@ export async function runKalign(srcCol: DG.Column, isAligned = false, unUsedName
     sequences = sequences.map((v: string, _) => AlignedSequenceEncoder.clean(v).replace(/\-/g, ''));
 
   const fasta = _stringsToFasta(sequences);
-  const CLI = await new Aioli({
-    tool: 'kalign',
-    version: '3.3.1',
-    reinit: true,
-  });
+  const CLI = await new Aioli([
+    'base/1.0.0',
+    {tool: 'kalign', version: '3.3.1', reinit: true,}
+  ]);
 
   console.log(['fasta.length =', fasta.length]);
 

package/src/utils/save-as-fasta.ts

@@ -0,0 +1,109 @@
+import * as DG from 'datagrok-api/dg';
+import * as ui from 'datagrok-api/ui';
+import * as grok from 'datagrok-api/grok';
+import {SplitterFunc, WebLogo} from '@datagrok-libraries/bio/src/viewers/web-logo';
+import wu from 'wu';
+import {UnitsHandler} from '@datagrok-libraries/bio/src/utils/units-handler';
+
+const FASTA_LINE_WIDTH = 60;
+
+/** Shows dialog to select id columns list and seq column, builds and downloads FASTA content */
+export function saveAsFastaUI() {
+  // Use grid for column order adjusted by user
+  let grid: DG.Grid = grok.shell.tv.grid;
+
+  const idGColList: DG.GridColumn[] = wu.count(0).take(grid.columns.length)
+    .map((colI: number) => grid.columns.byIndex(colI)!)
+    .filter((gcol: DG.GridColumn) => gcol.column ? gcol.column.semType !== DG.SEMTYPE.MACROMOLECULE : false).toArray();
+  const defaultIdGCol: DG.GridColumn | undefined = idGColList
+    .find((gcol: DG.GridColumn) => gcol.name.toLowerCase().indexOf('id') !== -1);
+  const idDefaultValue = defaultIdGCol ? [defaultIdGCol.name] : [];
+
+  const idGColListInput = ui.multiChoiceInput('Seq id columns', idDefaultValue,
+    idGColList.map((gcol: DG.GridColumn) => gcol.name));
+
+  const seqGColList: DG.GridColumn[] = wu.count(0).take(grid.columns.length)/* range rom 0 to grid.columns.length */
+    .map((colI: number) => grid.columns.byIndex(colI)!)
+    .filter((gc: DG.GridColumn) => {
+      const col: DG.Column | null = gc.column;
+      if (col && col.semType === DG.SEMTYPE.MACROMOLECULE) {
+        const uh = new UnitsHandler(col);
+        return uh.isFasta();
+      }
+      return false;
+    }).toArray();
+
+  const seqDefaultValue = seqGColList.length > 0 ? seqGColList[0].name : [];
+  const seqColInput = ui.choiceInput('Seq column', seqDefaultValue,
+    seqGColList.map((gCol: DG.GridColumn) => gCol.name));
+
+  const lineWidthInput = ui.intInput('FASTA line width', FASTA_LINE_WIDTH);
+
+  ui.dialog({title: 'Save as FASTA',})
+    .add(ui.inputs([
+      idGColListInput,
+      seqColInput,
+      lineWidthInput
+    ]))
+    .onOK(() => {
+      const valueIdColList: DG.Column[] = idGColListInput.value ?
+        idGColListInput.value.map((colName: string) => grid.columns.byName(colName)!.column!) : [];
+      const valueSeqCol: DG.Column | null = seqColInput.value ?
+        grid.columns.byName(seqColInput.value as string)!.column : null;
+      const valueLineWidth = lineWidthInput.value ?? FASTA_LINE_WIDTH;
+
+      if (!valueSeqCol)
+        grok.shell.warning(`Seq column is mandatory to save as FASTA.`);
+
+      const resFastaTxt: string = saveAsFastaDo(valueIdColList, valueSeqCol!, valueLineWidth);
+
+      const aEl: HTMLAnchorElement = document.createElement('a',);
+      aEl.setAttribute('href', `data:text/plain;charset=utf-8,${encodeURIComponent(resFastaTxt)}`);
+      aEl.setAttribute('download', `${grid.dataFrame.name}.fasta`);
+      aEl.click();
+    })
+    .show();
+}
+
+/** */
+export function saveAsFastaDo(
+  idColList: DG.Column[], seqCol: DG.Column, lineWidth: number = FASTA_LINE_WIDTH, lineSeparator: string = '\n'
+): string {
+  const splitter: SplitterFunc = WebLogo.splitterAsFasta;
+
+  const fastaLines: string[] = [];
+
+  for (let rowI: number = 0; rowI < seqCol.length; rowI++) {
+    // multiple identifiers separated by vertical bars
+    // https://en.wikipedia.org/wiki/FASTA_format
+
+    const seqId: string = idColList.map((col) => col.get(rowI).toString()).join('|');
+    const seq: string = seqCol.get(rowI);
+    const seqLineList: string[] = wrapSequence(seq, splitter, lineWidth);
+
+    fastaLines.push(`>${seqId}${lineSeparator}`);
+    for (const line of seqLineList)
+      fastaLines.push(`${line}${lineSeparator}`);
+  }
+
+  //return fastaLines.join(lineSeparator);
+  return ''.concat(...fastaLines);
+}
+
+/* split sequence for monomers to prevent wrapping monomer partially */
+export function wrapSequence(seq: string, splitter: SplitterFunc, lineWidth: number = FASTA_LINE_WIDTH): string[] {
+  const seqMonomerList = splitter(seq);
+  let seqPos: number = 0;
+  const seqLength: number = seqMonomerList.length;
+
+  const seqLineList: string[] = [];
+  while (seqPos < seqLength) {
+    /* join sliced monomer into line */
+    const seqLine: string[] = seqMonomerList.slice(seqPos, seqPos + lineWidth);
+    const seqLineTxt: string = seqLine.map((m) => m.length > 1 ? `[${m}]` : m).join('');
+    seqLineList.push(seqLineTxt);
+    seqPos += seqLine.length;
+  }
+
+  return seqLineList;
+}

package/src/viewers/vd-regions-viewer.ts

@@ -113,8 +113,10 @@ export class VdRegionsViewer extends DG.JsViewer implements IVdRegionsViewer {
     await this.buildView();
   }
 
-  public override async onTableAttached() {
-    await this.init();
+  public override onTableAttached() {
+    window.setTimeout(async () => {
+      await this.init();
+    }, 0 /* next event cycle */);
   }
 
   public override async onPropertyChanged(property: DG.Property | null) {