@datagrok/bio 2.10.22 → 2.10.24

package/files/samples/HELM-non-linear.csv

@@ -0,0 +1,8 @@
+HELM
+"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]}$PEPTIDE1,PEPTIDE1,16:R2-1:R1$$$V2.0"
+PEPTIDE1{meI.hHis.Aca.Cys_SEt.T.dK.Thr_PO3H2.Aca.Tyr_PO3H2.D-Chg.dV.Phe_ab-dehydro.N.D-Orn.D-aThr.Phe_4Me}$$$$
+"PEPTIDE1{C.A.C.A.C.A.C.A}|PEPTIDE2{C.A.C.A.C.A.C.A}|PEPTIDE3{C.A.C.A.C.A.C.A}|PEPTIDE4{C.A.C.A.C.A.C.A}$PEPTIDE1,PEPTIDE2,3:R3-8:R2|PEPTIDE3,PEPTIDE4,3:R3-8:R2|PEPTIDE1,PEPTIDE3,5:R3-5:R3$$$V2.0"
+"PEPTIDE1{C.A.C.A.C.A.C.A.C.A.C.A}|PEPTIDE2{C.A.C.A.C.A.C.A.C.A.C.A}$PEPTIDE2,PEPTIDE1,1:R1-3:R3|PEPTIDE2,PEPTIDE2,1:R3-12:R2$$$V2.0"
+RNA1{d(A)p.d(G)p.d(C)p.d(T)p.d(A)p.d(G)p.d(C)p.d(T)p}$$$$V2.0
+RNA1{d(A)p.d(G)p.d(C)p.d(T)p.d(A)p.d(G)p.d(C)p.d(T)p}|RNA2{d(G)p.d(G)p.d(G)p.d(G)p.d(A)p.d(G)p.d(C)p.d(T)p}$$$$V2.0
+"RNA1{d(A)p.d(G)p.d(C)p.d(T)p.d(A)p.d(G)p.d(C)p.d(T)p}$RNA1,RNA1,24:R2-1:R1$$$V2.0"

package/package.json

@@ -5,7 +5,7 @@
     "name": "Leonid Stolbov",
     "email": "lstolbov@datagrok.ai"
   },
-  "version": "2.10.22",
+  "version": "2.10.24",
   "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.",
   "repository": {
     "type": "git",
@@ -34,7 +34,7 @@
   ],
   "dependencies": {
     "@biowasm/aioli": "^3.1.0",
-    "@datagrok-libraries/bio": "^5.38.11",
+    "@datagrok-libraries/bio": "^5.38.12",
     "@datagrok-libraries/chem-meta": "^1.0.1",
     "@datagrok-libraries/ml": "^6.3.49",
     "@datagrok-libraries/tutorials": "^1.3.6",

package/src/demo/bio03-atomic-level.ts

@@ -35,7 +35,7 @@ export async function demoBio03UI(): Promise<void> {
       })
       .step('To atomic level', async () => {
         const seqCol = df.getCol(seqColName);
-        await toAtomicLevel(df, seqCol);
+        await toAtomicLevel(df, seqCol, false);
       }, {
         description: 'Get atomic level structures of Macromolecules.',
         delay: 2000,

package/src/package.ts

@@ -60,7 +60,7 @@ import {SplitToMonomersFunctionEditor} from './function-edtiors/split-to-monomer
 import {splitToMonomersUI} from './utils/split-to-monomers';
 import {MonomerCellRenderer} from './utils/monomer-cell-renderer';
 import {BioPackage, BioPackageProperties} from './package-types';
-import {RDModule} from '@datagrok-libraries/chem-meta/src/rdkit-api';
+import {RDMol} from '@datagrok-libraries/chem-meta/src/rdkit-api';
 import {PackageSettingsEditorWidget} from './widgets/package-settings-editor-widget';
 import {getCompositionAnalysisWidget} from './widgets/composition-analysis-widget';
 import {MacromoleculeColumnWidget} from './utils/macromolecule-column-widget';
@@ -69,9 +69,11 @@ import {_getEnumeratorWidget, _setPeptideColumn} from './utils/enumerator-tools'
 import {getRegionDo} from './utils/get-region';
 import {GetRegionApp} from './apps/get-region-app';
 import {GetRegionFuncEditor} from './utils/get-region-func-editor';
+import {HelmToMolfileConverter} from './utils/helm-to-molfile';
 import {DIMENSIONALITY_REDUCER_TERMINATE_EVENT}
   from '@datagrok-libraries/ml/src/workers/dimensionality-reducing-worker-creator';
 import {Options} from '@datagrok-libraries/utils/src/type-declarations';
+import { sequenceToMolfile } from './utils/sequence-to-mol';
 
 export const _package = new BioPackage();
 
@@ -385,7 +387,7 @@ export async function getRegionTopMenu(
 //input: object options {optional: true}
 //output: viewer result
 //editor: Bio:SeqActivityCliffsEditor
-export async function activityCliffs(df: DG.DataFrame, macroMolecule: DG.Column, activities: DG.Column,
+export async function activityCliffs(df: DG.DataFrame, macroMolecule: DG.Column<string>, activities: DG.Column,
   similarity: number, methodName: DimReductionMethods, options?: (IUMAPOptions | ITSNEOptions) & Options,
 ): Promise<DG.Viewer | undefined> {
   if (!checkInputColumnUI(macroMolecule, 'Activity Cliffs'))
@@ -495,8 +497,8 @@ export async function sequenceSpaceTopMenu(
         embedYCol = table.columns.byName(embedColsNames[1]);
       }
 
-      embedXCol.init((i) => embeddings[i][0]);
-      embedYCol.init((i) => embeddings[i][1]);
+      embedXCol.init((i) => embeddings[i] ? embeddings[i][0] : undefined);
+      embedYCol.init((i) => embeddings[i] ? embeddings[i][1] : undefined);
       const progress = (_nEpoch / epochsLength * 100);
       pg.update(progress, `Running sequence space ... ${progress.toFixed(0)}%`);
     }
@@ -613,29 +615,18 @@ export async function sequenceSpaceTopMenu(
         sp = (v as DG.TableView).scatterPlot({x: embedColsNames[0], y: embedColsNames[1], title: 'Sequence space'});
     }
   } */
-};
+}
 
 //top-menu: Bio | Convert | To Atomic Level...
 //name: To Atomic Level
 //description: Converts sequences to molblocks
 //input: dataframe df [Input data table]
 //input: column macroMolecule {semType: Macromolecule}
-export async function toAtomicLevel(df: DG.DataFrame, macroMolecule: DG.Column): Promise<void> {
-  if (DG.Func.find({package: 'Chem', name: 'getRdKitModule'}).length === 0) {
-    grok.shell.warning('Transformation to atomic level requires package "Chem" installed.');
-    return;
-  }
-  if (!checkInputColumnUI(macroMolecule, 'To Atomic Level'))
-    return;
-  const monomerLib: IMonomerLib = (await getMonomerLibHelper()).getBioLib();
-  const atomicLevelRes = await _toAtomicLevel(df, macroMolecule, monomerLib);
-  if (atomicLevelRes.col !== null) {
-    df.columns.add(atomicLevelRes.col, true);
-    await grok.data.detectSemanticTypes(df);
-  }
-
-  if (atomicLevelRes.warnings && atomicLevelRes.warnings.length > 0)
-    grok.shell.warning(ui.list(atomicLevelRes.warnings));
+//input: bool nonlinear=false { description: Slower mode for cycling/branching HELM structures }
+export async function toAtomicLevel(df: DG.DataFrame, macroMolecule: DG.Column, nonlinear: boolean): Promise<void> {
+  const pi = DG.TaskBarProgressIndicator.create('Converting to atomic level ...');
+  sequenceToMolfile(df, macroMolecule, nonlinear);
+  pi.close();
 }
 
 //top-menu: Bio | Analyze | MSA...
@@ -1059,6 +1050,7 @@ export function getEnumeratorWidget(molColumn: DG.Column): DG.Widget {
   return _getEnumeratorWidget(molColumn);
 }
 
+
 //top-menu: Bio | Convert | SDF to JSON Library...
 //name: SDF to JSON Library
 //input: dataframe table

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

@@ -1,7 +1,7 @@
 import * as grok from 'datagrok-api/grok';
 import * as DG from 'datagrok-api/dg';
 
-import {category, expect, expectArray, test, awaitCheck, delay} from '@datagrok-libraries/utils/src/test';
+import {category, expect, expectArray, test, testEvent} from '@datagrok-libraries/utils/src/test';
 import {ALPHABET, NOTATION, TAGS as bioTAGS} from '@datagrok-libraries/bio/src/utils/macromolecule';
 import {
   countForMonomerAtPosition,
@@ -30,9 +30,9 @@ ATC-G-TTGC--
     seqCol.setTag(bioTAGS.aligned, 'SEQ.MSA');
 
     const wlViewer: WebLogoViewer = (await df.plot.fromType('WebLogo')) as WebLogoViewer;
-    tv.dockManager.dock(wlViewer.root, DG.DOCK_TYPE.DOWN);
-
-    await awaitCheck(() => wlViewer.Length > 0, 'WebLogo.Length is zero', 100);
+    await testEvent(wlViewer.onLayoutCalculated, () => {}, () => {
+      tv.dockManager.dock(wlViewer.root, DG.DOCK_TYPE.DOWN);
+    }, 200);
     const positions: PI[] = wlViewer['positions'];
 
     const resAllDf1: PI[] = [
@@ -82,9 +82,9 @@ ATC-G-TTGC--
     df.filter.fireChanged();
     const wlViewer: WebLogoViewer = (await df.plot.fromType('WebLogo',
       {'shrinkEmptyTail': true})) as WebLogoViewer;
-    tv.dockManager.dock(wlViewer.root, DG.DOCK_TYPE.DOWN);
-
-    await awaitCheck(() => wlViewer.Length > 0, 'WebLogo.Length is zero', 100);
+    await testEvent(wlViewer.onLayoutCalculated, () => {}, () => {
+      tv.dockManager.dock(wlViewer.root, DG.DOCK_TYPE.DOWN);
+    }, 200);
     const positions: PI[] = wlViewer['positions'];
 
     const resAllDf1: PI[] = [
@@ -120,9 +120,9 @@ ATC-G-TTGC--
 
     const wlViewer: WebLogoViewer = (await df.plot.fromType('WebLogo',
       {'skipEmptyPositions': true})) as WebLogoViewer;
-    tv.dockManager.dock(wlViewer.root, DG.DOCK_TYPE.DOWN);
-
-    await awaitCheck(() => wlViewer.Length > 0, 'WebLogo.Length is zero');
+    await testEvent(wlViewer.onLayoutCalculated, () => {}, () => {
+      tv.dockManager.dock(wlViewer.root, DG.DOCK_TYPE.DOWN);
+    }, 200);
     const resPosList: PI[] = wlViewer['positions'];
 
     const tgtPosList: PI[] = [
@@ -156,9 +156,9 @@ ATC-G-TTGC--
       endPositionName: '7',
       skipEmptyPositions: true,
     })) as WebLogoViewer;
-    tv.dockManager.dock(wlViewer.root, DG.DOCK_TYPE.DOWN);
-
-    await awaitCheck(() => wlViewer.Length > 0, 'WebLogo.Length is zero', 100);
+    await testEvent(wlViewer.onLayoutCalculated, () => {}, () => {
+      tv.dockManager.dock(wlViewer.root, DG.DOCK_TYPE.DOWN);
+    }, 200);
     const resPosList: PI[] = wlViewer['positions'];
     const tgtPosList: PI[] = [
       new PI(2, '3', {'C': new PMI(5)}),

package/src/tests/detectors-benchmark-tests.ts

@@ -29,7 +29,7 @@ category('detectorsBenchmark', () => {
   });
 
   test('fastaDnaLong1e6Few50', async () => {
-    await detectMacromoleculeBenchmark(15, NOTATION.FASTA, ALPHABET.DNA, 1E6, 50);
+    await detectMacromoleculeBenchmark(20, NOTATION.FASTA, ALPHABET.DNA, 1E6, 50);
   });
 
   // -- separator --
@@ -49,7 +49,7 @@ category('detectorsBenchmark', () => {
   async function detectMacromoleculeBenchmark(
     maxET: number, notation: NOTATION, alphabet: ALPHABET, length: number, count: number, separator?: string,
   ): Promise<number> {
-    return await benchmark<DG.FuncCall, DG.Column>(10,
+    return await benchmark<DG.FuncCall, DG.Column>(maxET,
       (): DG.FuncCall => {
         const col: DG.Column = generate(notation, [...getAlphabet(alphabet)], length, count, separator);
         const funcCall: DG.FuncCall = detectFunc.prepare({col: col});

package/src/tests/detectors-tests.ts

@@ -182,7 +182,10 @@ MWRSWY-CKHP`;
           const df: DG.DataFrame = await readFile(samples[key]);
           // await grok.data.detectSemanticTypes(df);
           return df;
-        })();
+        })().catch((err: any) => {
+          delete _samplesDfs[key];
+          throw err;
+        });
       }
       return _samplesDfs[key];
     };

package/src/tests/to-atomic-level-tests.ts

@@ -64,7 +64,7 @@ category('toAtomicLevel', async () => {
 
   async function getTestResult(source: DG.DataFrame, target: DG.DataFrame): Promise<void> {
     const inputCol = source.getCol(inputColName);
-    await toAtomicLevel(source, inputCol);
+    await toAtomicLevel(source, inputCol, false);
     const obtainedCol = source.getCol(outputColName);
     const expectedCol = target.getCol(outputColName);
     const obtainedArray = [...obtainedCol.values()];

package/src/utils/cell-renderer.ts

@@ -27,6 +27,7 @@ import * as C from './constants';
 import {_package, getBioLib} from '../package';
 import {ISeqSplitted} from '@datagrok-libraries/bio/src/utils/macromolecule/types';
 import {getSplitter} from '@datagrok-libraries/bio/src/utils/macromolecule/utils';
+import {errInfo} from './err-info';
 
 
 type TempType = { [tagName: string]: any };
@@ -136,7 +137,12 @@ export class MacromoleculeSequenceCellRenderer extends DG.GridCellRenderer {
 
     // TODO: Store temp data to GridColumn
     // Now the renderer requires data frame table Column underlying GridColumn
-    const grid = gridCell.grid;
+    let grid: DG.Grid | undefined = undefined;
+    try { grid = gridCell.grid; } catch (err: any) {
+      grid = undefined;
+      const [errMsg, errStack] = errInfo(err);
+      _package.logger.error(errMsg, undefined, errStack);
+    }
     const tableCol: DG.Column = gridCell.cell.column;
     if (!grid || !tableCol) return;
 

package/src/utils/err-info.ts

@@ -0,0 +1,28 @@
+import * as grok from 'datagrok-api/grok';
+import * as ui from 'datagrok-api/ui';
+import * as DG from 'datagrok-api/dg';
+
+export function errMsg(err: any): string {
+  if (typeof err === 'string' || err instanceof String)
+    return err as string;
+  else if (err.constructor.name === 'StateError')
+    return err['message'];
+  else if (err.constructor.name === 'StateError' && '$thrownJsError' in err)
+    return errMsg(err['$thrownJsError']);
+  else if (err instanceof Error)
+    return (err as Error).message;
+  else
+    return err.toString();
+}
+
+export function errStack(err: any): string | undefined {
+  if (err instanceof Error)
+    return err.stack;
+  else if (err.constructor.name === 'StateError' && '$thrownJsError' in err)
+    return errStack(err['$thrownJsError']);
+  return undefined;
+}
+
+export function errInfo(err: any): [string, string | undefined] {
+  return [errMsg(err), errStack(err)];
+}