@datagrok/bio 2.0.21 → 2.0.23

package/package.json

@@ -5,7 +5,7 @@
     "name": "Leonid Stolbov",
     "email": "lstolbov@datagrok.ai"
   },
-  "version": "2.0.21",
+  "version": "2.0.23",
   "description": "Bio is a [package](https://datagrok.ai/help/develop/develop#packages) for the [Datagrok](https://datagrok.ai) platform",
   "repository": {
     "type": "git",
@@ -17,7 +17,7 @@
     "@datagrok-libraries/bio": "^5.1.1",
     "@datagrok-libraries/chem-meta": "1.0.1",
     "@datagrok-libraries/ml": "^6.2.0",
-    "@datagrok-libraries/utils": "^1.10.1",
+    "@datagrok-libraries/utils": "^1.11.0",
     "@deck.gl/core": "^8.7.5",
     "@deck.gl/layers": "^8.7.5",
     "@luma.gl/constants": "^8.5.10",

package/src/package-test.ts

@@ -24,8 +24,9 @@ export {tests};
 //name: test
 //input: string category {optional: true}
 //input: string test {optional: true}
+//input: bool catchUnhandled {optional: true}
 //output: dataframe result
-export async function test(category: string, test: string): Promise<DG.DataFrame> {
-  const data = await runTests({category, test});
+export async function test(category: string, test: string, catchUnhandled: boolean): Promise<DG.DataFrame> {
+  const data = await runTests({category, test, catchUnhandled});
   return DG.DataFrame.fromObjects(data)!;
 }

package/src/package.ts

@@ -99,7 +99,7 @@ export function checkInputColumn(
     ) {
       const notationAdd = allowedNotations.length == 0 ? 'any notation' :
         (`notation${allowedNotations.length > 1 ? 's' : ''} ${allowedNotations.map((n) => `"${n}"`).join(', ')} `);
-      msg = `${name} + ' analysis is allowed for Macromolecules with notation ${notationAdd}.`;
+      msg = `${name} analysis is allowed for Macromolecules with ${notationAdd}.`;
       res = false;
     } else if (!uh.isHelm()) {
       // alphabet is not specified for 'helm' notation
@@ -521,7 +521,7 @@ export function saveAsFasta() {
 }
 
 //name: BioSubstructureFilter
-//description: Substructure filter for linear macromolecules
+//description: Substructure filter for macromolecules
 //tags: filter
 //output: filter result
 //meta.semType: Macromolecule

package/src/substructure-search/substructure-search.ts

@@ -78,7 +78,7 @@ export function linearSubstructureSearch(substructure: string, col: DG.Column):
   return resultArray;
 }
 
-async function helmSubstructureSearch(substructure: string, col: DG.Column): Promise<BitSet> {
+export async function helmSubstructureSearch(substructure: string, col: DG.Column): Promise<BitSet> {
   const helmColWithSubstructure = DG.Column.string('helm', col.length + 1)
     .init((i) => i === col.length ? substructure : col.get(i));
   helmColWithSubstructure.setTag(DG.TAGS.UNITS, bio.NOTATION.HELM);

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

@@ -38,6 +38,7 @@ ATC-G-TTGC--
     seqCol.semType = DG.SEMTYPE.MACROMOLECULE;
     seqCol.setTag(DG.TAGS.UNITS, bio.NOTATION.FASTA);
     seqCol.setTag(bio.TAGS.alphabet, bio.ALPHABET.DNA);
+    seqCol.setTag(bio.TAGS.aligned, 'SEQ.MSA');
 
     const wlViewer: bio.WebLogo = (await df.plot.fromType('WebLogo')) as bio.WebLogo;
     tv.dockManager.dock(wlViewer.root, DG.DOCK_TYPE.DOWN);
@@ -74,12 +75,12 @@ ATC-G-TTGC--
 
   test('positions with shrinkEmptyTail option true (filterd)', async () => {
     let csvDf2 = `seq 
-    -TC-G-TTGC--
-    -TC-GCTTGC--
-    -T--C-GT-
-    -T--C-GT-
-    -T--C-GT-
-    -T--CCGT-`;
+-TC-G-TTGC--
+-TC-GCTTGC--
+-T--C-GT-
+-T--C-GT-
+-T--C-GT-
+-T--CCGT-`;
     const df: DG.DataFrame = DG.DataFrame.fromCsv(csvDf2);
     const tv: DG.TableView = grok.shell.addTableView(df);
 
@@ -87,6 +88,7 @@ ATC-G-TTGC--
     seqCol.semType = DG.SEMTYPE.MACROMOLECULE;
     seqCol.setTag(DG.TAGS.UNITS, bio.NOTATION.FASTA);
     seqCol.setTag(bio.TAGS.alphabet, bio.ALPHABET.DNA);
+    seqCol.setTag(bio.TAGS.aligned, 'SEQ');
 
     df.filter.init((i) => {
       return i > 2;
@@ -131,6 +133,7 @@ ATC-G-TTGC--
     seqCol.semType = DG.SEMTYPE.MACROMOLECULE;
     seqCol.setTag(DG.TAGS.UNITS, bio.NOTATION.FASTA);
     seqCol.setTag(bio.TAGS.alphabet, bio.ALPHABET.DNA);
+    seqCol.setTag(bio.TAGS.aligned, 'SEQ.MSA');
 
     const wlViewer: bio.WebLogo = (await df.plot.fromType('WebLogo',
       {'skipEmptyPositions': true})) as bio.WebLogo;

package/src/tests/checkInputColumn-tests.ts

@@ -27,11 +27,13 @@ seq4`;
     const df: DG.DataFrame = DG.DataFrame.fromCsv(csv);
     const col: DG.Column = df.getCol('seq');
     col.semType = DG.SEMTYPE.MACROMOLECULE;
-    col.setTag(DG.TAGS.UNITS, 'fasta');
+    col.setTag(DG.TAGS.UNITS, bio.NOTATION.FASTA);
     col.setTag(bio.TAGS.alphabet, bio.ALPHABET.DNA);
+    col.setTag(bio.TAGS.aligned, 'SEQ');
 
     const [res, msg]: [boolean, string] = checkInputColumn(
-      col, 'Test', ['fasta',], ['DNA', 'RNA', 'PT']);
+      col, 'Test', [bio.NOTATION.FASTA,],
+      [bio.ALPHABET.DNA, bio.ALPHABET.RNA, bio.ALPHABET.PT]);
 
     expect(res, true);
   });
@@ -45,7 +47,8 @@ seq4`;
     col.setTag(bio.TAGS.alphabetIsMultichar, 'true');
 
     const [res, msg]: [boolean, string] = checkInputColumn(
-      col, 'Test', ['fasta',], ['DNA', 'RNA', 'PT']);
+      col, 'Test', [bio.NOTATION.FASTA,],
+      [bio.ALPHABET.DNA, bio.ALPHABET.RNA, bio.ALPHABET.PT]);
 
     expect(res, false);
   });
@@ -58,9 +61,11 @@ seq4`;
     col.setTag(bio.TAGS.alphabet, 'UN');
     col.setTag(bio.TAGS.alphabetSize, '11');
     col.setTag(bio.TAGS.alphabetIsMultichar, 'true');
+    col.setTag(bio.TAGS.aligned, 'SEQ');
 
     const [res, msg]: [boolean, string] = checkInputColumn(
-      col, 'Test', ['fasta',], ['DNA', 'RNA', 'PT']);
+      col, 'Test', [bio.NOTATION.FASTA,],
+      [bio.ALPHABET.DNA, bio.ALPHABET.RNA, bio.ALPHABET.PT]);
 
     expect(res, false);
   });

package/src/widgets/bio-substructure-filter.ts

@@ -6,18 +6,21 @@
 
 import * as ui from 'datagrok-api/ui';
 import * as DG from 'datagrok-api/dg';
-import * as bio from '@datagrok-libraries/bio';
 
+import * as grok from 'datagrok-api/grok';
 import wu from 'wu';
-import {linearSubstructureSearch} from '../substructure-search/substructure-search';
+import {helmSubstructureSearch, linearSubstructureSearch} from '../substructure-search/substructure-search';
 import {Subject, Subscription} from 'rxjs';
 import * as C from '../utils/constants';
+import {updateDivInnerHTML} from '../utils/ui-utils';
+import {NOTATION} from '@datagrok-libraries/bio';
 
 export class BioSubstructureFilter extends DG.Filter {
-  bioFilter: FastaFilter | SeparatorFilter | null = null;
+  bioFilter: FastaFilter | SeparatorFilter | HelmFilter | null = null;
   bitset: DG.BitSet | null = null;
   loader: HTMLDivElement = ui.loader();
   onBioFilterChangedSubs?: Subscription;
+  notation: string | undefined = '';
 
   get calculating(): boolean { return this.loader.style.display == 'initial'; }
 
@@ -45,9 +48,10 @@ export class BioSubstructureFilter extends DG.Filter {
     super.attach(dataFrame);
     this.column = dataFrame.columns.bySemType(DG.SEMTYPE.MACROMOLECULE);
     this.columnName = this.column?.name;
-    const notation = this.column?.getTag(DG.TAGS.UNITS);
-    this.bioFilter = notation === bio.NOTATION.FASTA ?
-      new FastaFilter() : new SeparatorFilter(this.column!.getTag(C.TAGS.SEPARATOR));
+    this.notation = this.column?.getTag(DG.TAGS.UNITS);
+    this.bioFilter = this.notation === NOTATION.FASTA ?
+      new FastaFilter() : this.notation === NOTATION.SEPARATOR ?
+        new SeparatorFilter(this.column!.getTag(C.TAGS.SEPARATOR)) : new HelmFilter();
     this.root.appendChild(this.bioFilter!.filterPanel);
     this.root.appendChild(this.loader);
 
@@ -76,7 +80,7 @@ export class BioSubstructureFilter extends DG.Filter {
   applyState(state: any): void {
     super.applyState(state);
     if (state.bioSubstructure)
-      this.bioFilter!.substructureInput.value = state.bioSubstructure;
+      this.bioFilter!.substructure = state.bioSubstructure;
 
     const that = this;
     if (state.bioSubstructure)
@@ -98,7 +102,9 @@ export class BioSubstructureFilter extends DG.Filter {
     } else {
       this.calculating = true;
       try {
-        this.bitset = linearSubstructureSearch(this.bioFilter!.substructure, this.column!);
+        this.bitset = this.notation === NOTATION.HELM ?
+          await helmSubstructureSearch(this.bioFilter!.substructure, this.column!) :
+          linearSubstructureSearch(this.bioFilter!.substructure, this.column!);
         this.calculating = false;
         this.dataFrame?.rows.requestFilter();
       } finally {
@@ -108,13 +114,28 @@ export class BioSubstructureFilter extends DG.Filter {
   }
 }
 
-class FastaFilter {
+abstract class BioFilterBase {
   onChanged: Subject<any> = new Subject<any>();
+
+  get filterPanel() {
+    return new HTMLElement();
+  }
+
+  get substructure() {
+    return '';
+  }
+
+  set substructure(s: string) {
+  }
+}
+
+class FastaFilter extends BioFilterBase {
   substructureInput: DG.InputBase<string> = ui.stringInput('', '', () => {
     this.onChanged.next();
   }, {placeholder: 'Substructure'});
 
   constructor() {
+    super();
   }
 
   get filterPanel() {
@@ -124,6 +145,10 @@ class FastaFilter {
   get substructure() {
     return this.substructureInput.value;
   }
+
+  set substructure(s: string) {
+    this.substructureInput.value = s;
+  }
 }
 
 class SeparatorFilter extends FastaFilter {
@@ -150,4 +175,76 @@ class SeparatorFilter extends FastaFilter {
       this.substructureInput.value.replaceAll(this.separatorInput.value, this.colSeparator) :
       this.substructureInput.value;
   }
+
+  set substructure(s: string) {
+    this.substructureInput.value = s;
+  }
+}
+
+class HelmFilter extends BioFilterBase {
+  helmEditor: any;
+  _filterPanel = ui.div('', {style: {cursor: 'pointer'}});
+  helmSubstructure = '';
+
+  constructor() {
+    super();
+    this.init();
+  }
+
+  async init() {
+    this.helmEditor = await grok.functions.call('HELM:helmWebEditor');
+    await ui.tools.waitForElementInDom(this._filterPanel);
+    this.updateFilterPanel();
+    this._filterPanel.addEventListener('click', (event: MouseEvent) => {
+      //@ts-ignore
+      ui.dialog({showHeader: false, showFooter: true})
+        .add(this.helmEditor.editorView)
+        .onOK(() => {
+          const helmString = this.helmEditor
+            .webEditor.canvas.getHelm(true).replace(/<\/span>/g, '').replace(/<span style='background:#bbf;'>/g, '');
+          this.updateFilterPanel(helmString);
+          this.helmSubstructure = helmString;
+          this.onChanged.next();
+        }).show({modal: true, fullScreen: true});
+    });
+    ui.onSizeChanged(this._filterPanel).subscribe((_) => {
+      const helmString = this.helmEditor
+        .webEditor.canvas.getHelm(true).replace(/<\/span>/g, '').replace(/<span style='background:#bbf;'>/g, '');
+      this.updateFilterPanel(helmString);
+    });
+  }
+
+  get filterPanel() {
+    return this._filterPanel;
+  }
+
+  get substructure() {
+    return this.helmSubstructure;
+  }
+
+  set substructure(s: string) {
+    this.helmEditor.editor.setHelm(s);
+  }
+
+  updateFilterPanel(helmString?: string) {
+    const width = this._filterPanel.parentElement!.clientWidth < 100 ? 100 :
+      this._filterPanel.parentElement!.clientWidth;
+    const height = width / 2;
+    if (!helmString) {
+      const editDivStyle = {style: {
+        width: `${width}px`,
+        height: `${height/2}px`,
+        textAlign: 'center',
+        verticalAlign: 'middle',
+        lineHeight: `${height/2}px`,
+        border: '1px solid #dbdcdf'
+      }};
+      const editDiv = ui.divText('Click to edit', editDivStyle);
+      updateDivInnerHTML(this._filterPanel, editDiv);
+    } else {
+      updateDivInnerHTML(this._filterPanel, this.helmEditor.host);
+      this.helmEditor.editor.setHelm(helmString);
+      this.helmEditor.resizeEditor(width, height);
+    }
+  }
 }