@datagrok/helm 3.0.0 → 3.0.2

package/files/samples/helm-showcase.csv

@@ -8,20 +8,18 @@ Modified residues (meL/meY),Peptide / modified,PEPTIDE1{[meL].A.G.[meY].T}$$$$
 Head-to-tail cyclic (5),Peptide / cyclic,"PEPTIDE1{A.C.D.E.F}$PEPTIDE1,PEPTIDE1,5:R2-1:R1$$$V2.0"
 Disulfide cyclic,Peptide / cyclic,"PEPTIDE1{A.C.D.E.F.C}$PEPTIDE1,PEPTIDE1,2:R3-6:R3$$$V2.0"
 Cyclic head-to-tail (32-mer),Peptide / cyclic,"PEPTIDE1{A.C.D.E.F.G.H.I.K.L.M.N.P.Q.R.S.T.V.W.Y.A.C.D.E.F.G.H.I.K.L.M.N}$PEPTIDE1,PEPTIDE1,32:R2-1:R1$$$V2.0"
-Inner side-chain cycle (R3-R3),Peptide / cyclic,"PEPTIDE1{A.C.D.E.F.G.H.I.K.L}$PEPTIDE1,PEPTIDE1,3:R3-7:R3$$$V2.0"
-Two side-chain bridges (bicyclic),Peptide / multicyclic,"PEPTIDE1{A.C.D.E.F.G.H.I.K.L.M.N}$PEPTIDE1,PEPTIDE1,2:R3-6:R3|PEPTIDE1,PEPTIDE1,7:R3-11:R3$$$V2.0"
+Inner side-chain cycle (R3-R3),Peptide / cyclic,"PEPTIDE1{A.C.D.E.F.G.C.I.K.L}$PEPTIDE1,PEPTIDE1,3:R3-7:R3$$$V2.0"
+Two side-chain bridges (bicyclic),Peptide / multicyclic,"PEPTIDE1{A.C.D.E.F.D.D.I.K.L.dC.N}$PEPTIDE1,PEPTIDE1,2:R3-6:R3|PEPTIDE1,PEPTIDE1,7:R3-11:R3$$$V2.0"
 "Stapled helix (i, i+4)",Peptide / cyclic,"PEPTIDE1{A.C.D.E.C.G.H.I}$PEPTIDE1,PEPTIDE1,2:R3-5:R3$$$V2.0"
 Branched tri-ring (R3-R3 bridges),Peptide / multicyclic,"PEPTIDE1{C.A.C.A.C.A}|PEPTIDE2{C.A.C.A.C.A}|PEPTIDE3{C.A.C.A.C.A}$PEPTIDE1,PEPTIDE2,1:R3-1:R3|PEPTIDE2,PEPTIDE3,3:R3-3:R3|PEPTIDE3,PEPTIDE1,5:R3-5:R3$$$V2.0"
-Disulfide-rich (insulin-like),Peptide / multicyclic,"PEPTIDE1{G.I.V.E.Q.C.C.T.S.I.C.S.L.Y.Q.L.E.N.Y.C.N}|PEPTIDE2{F.V.N.Q.H.L.C.G.S.H.L.V.E.A.L.Y.L.V.C.G.E.R.G.F.F.Y.T.P.K.T}$PEPTIDE1,PEPTIDE2,7:R3-7:R3|PEPTIDE1,PEPTIDE2,20:R3-19:R3|PEPTIDE1,PEPTIDE1,6:R3-11:R3$$$V2.0"
+Disulfide-rich (insulin-like),Peptide / multicyclic,"PEPTIDE1{G.I.V.E.Q.C.C.T.S.I.C.S.L.Y.Q.L.E.N.Y.C.N}|PEPTIDE2{F.V.N.Q.H.L.C.G.S.H.L.V.E.A.L.Y.L.V.C.G.E.R.G.F.F.Y.T.P}$PEPTIDE1,PEPTIDE2,7:R3-7:R3|PEPTIDE1,PEPTIDE2,20:R3-19:R3|PEPTIDE1,PEPTIDE1,6:R3-11:R3$$$V2.0"
 Two disconnected chains,Peptide / linear,PEPTIDE1{A.C}|PEPTIDE2{G.H}$$$$V2.0
-"Ambiguous position (A,G,C)",Peptide / ambiguity,"PEPTIDE1{(A,G,C)}$$$$"
 RNA tetramer (ribose),Nucleic acid / single,RNA1{r(A)p.r(C)p.r(G)p.r(U)p}$$$$V2.0
 DNA tetramer (deoxy),Nucleic acid / single,RNA1{d(A)p.d(C)p.d(G)p.d(T)p}$$$$V2.0
 RNA 30-mer,Nucleic acid / single,RNA1{r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p}$$$$V2.0
 2'-Fluoro modified ribose,Nucleic acid / modified,RNA1{[fl2r](A)p.[fl2r](C)p.r(G)p.r(U)p}$$$$V2.0
-Phosphorothioate (PS) backbone,Nucleic acid / modified,RNA1{m(G)[sP].m(A)[sP].m(C)[sP].m(U)[sP]}$$$$V2.0
-2'-OMe gapmer (PS wings),Nucleic acid / modified,RNA1{m(A)[sP].m(C)[sP].d(G)p.d(A)p.d(T)p.d(C)p.m(G)[sP].m(U)[sP]}$$$$V2.0
-Bracket repeat (rA)x3,Nucleic acid / modified,RNA1{(r(A)p)'3'}$$$$V2.0
+Phosphorothioate (PS) backbone,Nucleic acid / modified,RNA1{m(G)[cmp].m(A)[msp].m(C)[eop].m(U)[sp]}$$$$V2.0
+2'-OMe gapmer (PS wings),Nucleic acid / modified,RNA1{m(A)[sp].m(C)[sp].d(G)p.d(A)p.d(T)p.d(C)p.m(G)[sp].m(U)[sp]}$$$$V2.0
 Mini duplex (2bp),Nucleic acid / duplex,"RNA1{r(A)p.r(C)p}|RNA2{r(G)p.r(U)p}$RNA1,RNA2,2:pair-5:pair$$$V2.0"
 Blunt duplex (8bp),Nucleic acid / duplex,"RNA1{r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p}|RNA2{r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p}$RNA1,RNA2,2:pair-23:pair|RNA1,RNA2,5:pair-20:pair|RNA1,RNA2,8:pair-17:pair|RNA1,RNA2,11:pair-14:pair|RNA1,RNA2,14:pair-11:pair|RNA1,RNA2,17:pair-8:pair|RNA1,RNA2,20:pair-5:pair|RNA1,RNA2,23:pair-2:pair$$$V2.0"
 Blunt duplex (12bp),Nucleic acid / duplex,"RNA1{r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p}|RNA2{r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p}$RNA1,RNA2,2:pair-35:pair|RNA1,RNA2,5:pair-32:pair|RNA1,RNA2,8:pair-29:pair|RNA1,RNA2,11:pair-26:pair|RNA1,RNA2,14:pair-23:pair|RNA1,RNA2,17:pair-20:pair|RNA1,RNA2,20:pair-17:pair|RNA1,RNA2,23:pair-14:pair|RNA1,RNA2,26:pair-11:pair|RNA1,RNA2,29:pair-8:pair|RNA1,RNA2,32:pair-5:pair|RNA1,RNA2,35:pair-2:pair$$$V2.0"
@@ -48,9 +46,9 @@ Asymmetric duplex (21 sense / 19 anti),Nucleic acid / siRNA,"RNA1{r(A)p.r(C)p.r(
 Palindromic self-pairing duplex,Nucleic acid / duplex,"RNA1{r(G)p.r(C)p.r(G)p.r(C)p.r(G)p.r(C)p.r(G)p.r(C)p}|RNA2{r(G)p.r(C)p.r(G)p.r(C)p.r(G)p.r(C)p.r(G)p.r(C)p}$RNA1,RNA2,2:pair-23:pair|RNA1,RNA2,5:pair-20:pair|RNA1,RNA2,8:pair-17:pair|RNA1,RNA2,11:pair-14:pair|RNA1,RNA2,14:pair-11:pair|RNA1,RNA2,17:pair-8:pair|RNA1,RNA2,20:pair-5:pair|RNA1,RNA2,23:pair-2:pair$$$V2.0"
 DNA blunt duplex (10bp),Nucleic acid / duplex,"RNA1{d(A)p.d(C)p.d(G)p.d(T)p.d(A)p.d(C)p.d(G)p.d(T)p.d(A)p.d(C)p}|RNA2{d(G)p.d(T)p.d(A)p.d(C)p.d(G)p.d(T)p.d(A)p.d(C)p.d(G)p.d(T)p}$RNA1,RNA2,2:pair-29:pair|RNA1,RNA2,5:pair-26:pair|RNA1,RNA2,8:pair-23:pair|RNA1,RNA2,11:pair-20:pair|RNA1,RNA2,14:pair-17:pair|RNA1,RNA2,17:pair-14:pair|RNA1,RNA2,20:pair-11:pair|RNA1,RNA2,23:pair-8:pair|RNA1,RNA2,26:pair-5:pair|RNA1,RNA2,29:pair-2:pair$$$V2.0"
 Tricyclic peptide (3 disulfide bridges),Peptide / multicyclic,"PEPTIDE1{C.A.C.A.C.A.C.A.C.A.C.A}$PEPTIDE1,PEPTIDE1,1:R3-5:R3|PEPTIDE1,PEPTIDE1,3:R3-9:R3|PEPTIDE1,PEPTIDE1,7:R3-11:R3$$$V2.0"
-Lasso peptide (macrolactam + tail),Peptide / cyclic,"PEPTIDE1{A.C.D.E.F.G.H.I}$PEPTIDE1,PEPTIDE1,4:R2-1:R1$$$V2.0"
+Lasso peptide (macrolactam + tail),Peptide / cyclic,"PEPTIDE1{A.C.D.E.F.G.H.I}$PEPTIDE1,PEPTIDE1,4:R3-1:R1$$$V2.0"
 RNA 40-mer single strand,Nucleic acid / single,RNA1{r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p}$$$$V2.0
 Two peptides bridged by one disulfide,Peptide / multicyclic,"PEPTIDE1{A.C.D.E}|PEPTIDE2{F.G.C.H}$PEPTIDE1,PEPTIDE2,2:R3-3:R3$$$V2.0"
 "siRNA 25-mer, 2-nt 3′ overhangs",Nucleic acid / siRNA,"RNA1{r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p}|RNA2{r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p.r(A)p.r(C)p.r(G)p.r(U)p}$RNA1,RNA2,2:pair-71:pair|RNA1,RNA2,5:pair-68:pair|RNA1,RNA2,8:pair-65:pair|RNA1,RNA2,11:pair-62:pair|RNA1,RNA2,14:pair-59:pair|RNA1,RNA2,17:pair-56:pair|RNA1,RNA2,20:pair-53:pair|RNA1,RNA2,23:pair-50:pair|RNA1,RNA2,26:pair-47:pair|RNA1,RNA2,29:pair-44:pair|RNA1,RNA2,32:pair-41:pair|RNA1,RNA2,35:pair-38:pair|RNA1,RNA2,38:pair-35:pair|RNA1,RNA2,41:pair-32:pair|RNA1,RNA2,44:pair-29:pair|RNA1,RNA2,47:pair-26:pair|RNA1,RNA2,50:pair-23:pair|RNA1,RNA2,53:pair-20:pair|RNA1,RNA2,56:pair-17:pair|RNA1,RNA2,59:pair-14:pair|RNA1,RNA2,62:pair-11:pair|RNA1,RNA2,65:pair-8:pair|RNA1,RNA2,68:pair-5:pair$$$V2.0"
 "Hairpin, long apical loop",Nucleic acid / structure,"RNA1{r(G)p.r(C)p.r(G)p.r(A)p.r(A)p.r(A)p.r(A)p.r(A)p.r(A)p.r(C)p.r(G)p.r(C)p}$RNA1,RNA1,2:pair-35:pair|RNA1,RNA1,5:pair-32:pair|RNA1,RNA1,8:pair-29:pair$$$V2.0"
-Peptide–PEG2–RNA triple conjugate,Conjugate / mixed,"PEPTIDE1{A.C.D.E.F}|RNA1{r(A)p.r(C)p.r(G)p}|CHEM1{[PEG2]}$PEPTIDE1,CHEM1,5:R2-1:R1|CHEM1,RNA1,1:R2-1:R1$$$V2.0"
+Peptide–PEG2–RNA triple conjugate,Conjugate / mixed,"PEPTIDE1{A.C.D.E.F}|RNA1{r(A)p.r(C)p.r(G)p}|CHEM1{[PEG2]}$PEPTIDE1,CHEM1,5:R2-1:R1|CHEM1,RNA1,1:R2-1:R1$$$V2.0"

package/package.json

@@ -1,31 +1,32 @@
 {
   "name": "@datagrok/helm",
   "friendlyName": "Helm",
-  "version": "3.0.0",
+  "version": "3.0.2",
   "author": {
-    "name": "Maria Dolotova",
-    "email": "mdolotova@datagrok.ai"
+    "name": "Davit Rizhinashvili",
+    "email": "drizhinashvili@datagrok.ai"
   },
   "description": "Provides support for HELM notation (importing, detecting, rendering, conversion).",
   "sources": [
     "css/helm.css"
   ],
   "dependencies": {
-    "@datagrok-libraries/bio": "^6.0.0",
+    "@datagrok-libraries/bio": "^6.0.2",
     "@datagrok-libraries/chem-meta": "^1.2.9",
-    "@datagrok-libraries/hwe": "^1.0.0",
+    "@datagrok-libraries/hwe": "^1.0.1",
+    "@datagrok-libraries/test": "^1.3.1",
     "@datagrok-libraries/utils": "^4.6.9",
     "cash-dom": "^8.1.1",
-    "datagrok-api": "^1.26.0",
+    "datagrok-api": "^1.27.3",
     "dayjs": "^1.10.6",
     "lru-cache": "^10.4.3",
     "rxjs": "^6.5.5",
-    "wu": "^2.1.0",
-    "@datagrok-libraries/test": "^1.1.0"
+    "wu": "^2.1.0"
   },
   "devDependencies": {
     "@datagrok/bio": "^2.25.1",
     "@datagrok/chem": "^1.17.0",
+    "@playwright/test": "^1.48.0",
     "@types/node": "17.0.45",
     "@types/webpack-env": "^1.18.5",
     "@types/wu": "^2.1.44",
@@ -78,6 +79,8 @@
   },
   "category": "Bioinformatics",
   "overrides": {
-    "datagrok-api": "$datagrok-api"
-  }
+    "datagrok-api": "$datagrok-api",
+    "uuid": "^11.1.1"
+  },
+  "playwrightTests": "playwright"
 }

package/src/helm-helper.ts

@@ -15,6 +15,9 @@ import type {RDModule} from '@datagrok-libraries/chem-meta/src/rdkit-api';
 
 import {HelmInput} from './widgets/helm-input';
 import {getHoveredMonomerFromEditorMol} from './utils/get-hovered';
+import * as DG from 'datagrok-api/dg';
+import * as grok from 'datagrok-api/grok';
+import * as ui from 'datagrok-api/ui';
 /* eslint-enable max-len */
 
 import {_package} from './package';
@@ -24,8 +27,10 @@ import {_package} from './package';
 // (`HelmHelperAdapter` over `HelmService`). The legacy forked Pistoia HELM Web
 // Editor (`org.helm.webeditor` / JSDraw2 / Dojo) is gone — this class no longer
 // touches any of those globals.
-import {HelmHelperAdapter, bridgeMonomerLib, HelmService} from '@datagrok-libraries/hwe';
+import {HelmHelperAdapter, bridgeMonomerLib, HelmService, PolymerTypes} from '@datagrok-libraries/hwe';
 import type {IMonomerLibBaseLike, MonomersFuncsLike} from '@datagrok-libraries/hwe';
+import {NOTATION} from '@datagrok-libraries/bio/src/utils/macromolecule';
+import {ISeqHandler} from '@datagrok-libraries/bio/src/utils/macromolecule/seq-handler';
 
 export class HelmHelper implements IHelmHelper {
   private static instanceCount: number = 0;
@@ -52,7 +57,15 @@ export class HelmHelper implements IHelmHelper {
   private get editorAdapter(): HelmHelperAdapter {
     if (this._editorAdapter === null) {
       const monomerLib = _package._libHelper!.getMonomerLib() as unknown as IMonomerLibBaseLike;
-      const service = new HelmService({monomerLib: bridgeMonomerLib(monomerLib), rdkitModule: this.rdKitModule});
+      const paneProviders: Map<string, (helm: string) => Promise<HTMLElement>> = new Map();
+      paneProviders.set('FASTA', async (helm) => this._toFastaProvider(helm));
+      paneProviders.set('BILN', async (helm) => this._toBilnProvider(helm));
+      paneProviders.set('Molecular Structure', async (helm) => this._atomicLevelProvider(helm));
+      paneProviders.set('Composition Analysis', async (helm) => this._compositionAnalysisProvider(helm));
+
+
+      const service = new HelmService({monomerLib: bridgeMonomerLib(monomerLib),
+        rdkitModule: this.rdKitModule, extraPanes: paneProviders});
       this._editorAdapter = new HelmHelperAdapter({service, seqHelper: this.seqHelper});
     }
     return this._editorAdapter;
@@ -127,4 +140,70 @@ export class HelmHelper implements IHelmHelper {
   removeGaps(srcHelm: string): HelmConvertRes {
     return this.editorAdapter.removeGaps(srcHelm);
   }
+
+  private _singeHelmDf: DG.DataFrame | null = null;
+  private _getSingleHelmDf(helm: string) {
+    if (this._singeHelmDf === null) {
+      const col = DG.Column.fromStrings('helm', [helm]);
+      col.setTag('quality', DG.SEMTYPE.MACROMOLECULE);
+      col.setTag('units', 'helm');
+      col.setTag('.alphabetIsMultichar', 'true');
+      col.setTag('alphabet', 'UN');
+      // set the width of widget in tag.
+      const ww = Math.floor(window.innerWidth * 0.6);
+      col.setTag('.toAtomicWidgetWidth', ww.toString());
+      this._singeHelmDf = DG.DataFrame.fromColumns([col]);
+    }
+    this._singeHelmDf.col('helm')!.set(0, helm);
+    return this._singeHelmDf;
+  }
+
+  private _singleHelmSh: ISeqHandler | null = null;
+  private _getSingleHelmDfSh(helm: string) {
+    if (!this._singleHelmSh) {
+      const df = this._getSingleHelmDf(helm);
+      this._singleHelmSh = this.seqHelper.getSeqHandler(df.col('helm')!);
+    }
+    return this._singleHelmSh;
+  }
+
+  private async _atomicLevelProvider(helm: string) {
+    const df = this._getSingleHelmDf(helm);
+    const semValue = DG.SemanticValue.fromTableCell(df.cell(0, 'helm')!);
+    const w: DG.Widget = await grok.functions.call('Bio:toAtomicLevelPanel', {sequence: semValue}) as DG.Widget;
+    return w.root;
+  }
+
+  private async _compositionAnalysisProvider(helm: string) {
+    const df = this._getSingleHelmDf(helm);
+    const semValue = DG.SemanticValue.fromTableCell(df.cell(0, 'helm')!);
+    const w: DG.Widget = await grok.functions.call('Bio:compositionAnalysisWidget', {sequence: semValue}) as DG.Widget;
+    return w.root;
+  }
+
+  private _toBilnProvider(helm: string) {
+    const parseResult = this.editorAdapter.service.parse(helm);
+    if (parseResult.mol.chains.length < 1 || parseResult.mol.chains.some((c) => c.polymerType !== PolymerTypes.PEPTIDE))
+      return ui.divText('Unsupported sequence for BILN');
+    const sh = this._getSingleHelmDfSh(helm);
+    const ta = ui.input.textArea('', {});
+    ta.value = sh.getConverter(NOTATION.BILN)(helm);
+    ta.input.setAttribute('disabled', 'true');
+    ta.input.style.width = '100%';
+    return ta.input;
+  }
+
+  private _toFastaProvider(helm: string) {
+    const parseResult = this.editorAdapter.service.parse(helm);
+    if (parseResult.mol.chains.length !== 1 ||
+      parseResult.mol.atoms.length !== parseResult.mol.bonds.filter((b) => b.kind !== 'hbond').length + 1
+    )
+      return ui.divText('Unsupported sequence for FASTA');
+    const sh = this._getSingleHelmDfSh(helm);
+    const ta = ui.input.textArea('', {});
+    ta.value = sh.getConverter(NOTATION.FASTA)(helm);
+    ta.input.setAttribute('disabled', 'true');
+    ta.input.style.width = '100%';
+    return ta.input;
+  }
 }

package/src/package-api.ts

@@ -41,6 +41,7 @@ export namespace funcs {
   }
 
   /**
+   * Convert a column of HELM macromolecules into a column of molfiles
    * @param {DG.Column} col
    *   semType: Macromolecule
    */

package/src/package-test.ts

@@ -1,10 +1,6 @@
 import * as DG from 'datagrok-api/dg';
 import * as grok from 'datagrok-api/grok';
 
-// Do not import anything from JsDrawLite or HelmWebEditor, only to the main Helm package
-// import {JSDraw2ModuleType} from '@datagrok/js-draw-lite/src/types/jsdraw2';
-// import {HelmType, OrgHelmModuleType} from '@datagrok/helm-web-editor/src/types/org-helm';
-
 import {runTests, tests, TestContext, initAutoTests as initTests} from '@datagrok-libraries/test/src/test';
 
 import './tests/_first-tests';

package/src/package.g.ts

@@ -43,6 +43,8 @@ export async function propertiesWidget(sequence: DG.SemanticValue) : Promise<any
   return await PackageFunctions.propertiesWidget(sequence);
 }
 
+//name: Get Molfiles from HELM
+//description: Convert a column of HELM macromolecules into a column of molfiles
 //input: column col { semType: Macromolecule }
 //output: column result
 export function getMolfiles(col: DG.Column<any>) : any {

package/src/package.ts

@@ -186,7 +186,10 @@ export class PackageFunctions {
     return getPropertiesWidget(sequence);
   }
 
-  @grok.decorators.func()
+  @grok.decorators.func({
+    'name': 'Get Molfiles from HELM',
+    'description': 'Convert a column of HELM macromolecules into a column of molfiles',
+  })
   static getMolfiles(
     @grok.decorators.param({'type': 'column', 'options': {'semType': 'Macromolecule'}}) col: DG.Column<string>): DG.Column<string> {
     const helmStrList = col.toList();