@datagrok/bio 2.26.2 → 2.26.4

package/package.json

@@ -5,7 +5,7 @@
     "name": "Davit Rizhinashvili",
     "email": "drizhinashvili@datagrok.ai"
   },
-  "version": "2.26.2",
+  "version": "2.26.4",
   "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",
@@ -68,7 +68,7 @@
   "devDependencies": {
     "@datagrok-libraries/helm-web-editor": "^1.1.16",
     "@datagrok-libraries/js-draw-lite": "^0.0.10",
-    "@datagrok/chem": "^1.15.0",
+    "@datagrok/chem": "^1.17.1",
     "@datagrok/dendrogram": "^1.2.33",
     "@datagrok/eda": "^1.4.13",
     "@datagrok/helm": "^2.13.1",

package/src/utils/cell-renderer.ts

@@ -216,11 +216,13 @@ export class MacromoleculeDifferenceCellRendererBack extends CellRendererWithMon
     const cell = gridCell.cell;
     const s: string = cell.value ?? '';
     const separator = this.tableCol.tags[bioTAGS.separator];
-    const units: string = this.tableCol.meta.units!;
+    let units: string = this.tableCol.meta.units!;
     w = getUpdatedWidth(grid, g, x, w, dpr);
     //TODO: can this be replaced/merged with splitSequence?
     const [s1, s2] = s.split('#');
-    const splitter = this.tableCol.temp[SeqTemps.notationProvider]?.separatorSplitter ?? getSplitter(units, separator);
+    if (units === NOTATION.CUSTOM && !this.tableCol.temp[SeqTemps.notationProvider])
+      units = NOTATION.SEPARATOR;
+    const splitter = this.tableCol.temp[SeqTemps.notationProvider]?.separatorSplitter ?? this.tableCol.temp[SeqTemps.notationProvider]?.splitter ?? getSplitter(units, separator);
     const s1SS = splitter(s1);
     const s2SS = splitter(s2);
     const subParts1 = wu.count(0).take(s1SS.length).map((posIdx) => s1SS.getCanonical(posIdx)).toArray();

package/src/utils/convert.ts

@@ -162,10 +162,16 @@ export async function convertDo(srcCol: DG.Column, seqHelper: ISeqHelper, target
         return '';
       }
     });
+    newCol.semType = DG.SEMTYPE.MACROMOLECULE;
+    newCol.meta.units = NOTATION.CUSTOM;
+    newCol.setTag('separator', '-');
+    newCol.setTag('aligned', 'SEQ');
+    newCol.setTag('alphabet', 'UN');
+    newCol.setTag('.alphabetIsMultichar', 'true');
     srcCol.dataFrame.columns.add(newCol);
-    const semType = await grok.functions.call('Bio:detectMacromolecule', {col: newCol});
-    if (semType)
-      newCol.semType = semType;
+    // const semType = await grok.functions.call('Bio:detectMacromolecule', {col: newCol});
+    // if (semType)
+    //   newCol.semType = semType;
     await grok.data.detectSemanticTypes(srcCol.dataFrame);
     return newCol;
   } else {

package/src/utils/seq-helper/seq-handler.ts

@@ -44,6 +44,15 @@ export class SeqHandler implements ISeqHandler {
   private _refinerPromise: Promise<void> = Promise.resolve();
   public get refinerPromise(): Promise<void> { return this._refinerPromise; }
 
+  private runInnerDetector() {
+    if (!this._column.temp['seqHandlerDetectorRun']) {
+      this._column.temp['seqHandlerDetectorRun'] = true;
+      const detectorFunc = DG.Func.find({name: 'detectMacromolecule', meta: {role: 'semTypeDetector'}})[0];
+      if (detectorFunc)
+        detectorFunc.applySync({col: this._column});
+    }
+  }
+
   protected constructor(col: DG.Column<string>,
     private readonly seqHelper: SeqHelper,
   ) {
@@ -53,19 +62,20 @@ export class SeqHandler implements ISeqHandler {
     let units: string | null = this._column.meta.units;
     if (!units) {
       // it may be from layout that the macromolecule semtype is set but every other tag is missing, so we manually run detectors
-      if (!this._column.temp['seqHandlerDetectorRun']) {
-        this._column.temp['seqHandlerDetectorRun'] = true;
-        const detectorFunc = DG.Func.find({name: 'detectMacromolecule', meta: {role: 'semTypeDetector'}})[0];
-        if (detectorFunc)
-          detectorFunc.applySync({col: this._column});
-        units = this._column.meta.units;
-      }
+      this.runInnerDetector();
+      units = this._column.meta.units;
       if (!units)
         throw new Error('Units are not specified in column');
     }
     this._units = units!;
 
     this._notation = this.getNotation();
+    if ([NOTATION.BILN, NOTATION.CUSTOM, NOTATION.SEPARATOR].includes(this._notation) && !this.column.getTag(TAGS.separator)) {
+      this.runInnerDetector();
+      if (!this.column.getTag(TAGS.separator))
+        throw new Error(`Separator tag '${TAGS.separator}' is not set for notation '${this._notation}'.`);
+    }
+
     if (this.isCustom() || this.isBiln()) {
       // this.column.temp[SeqTemps.notationProvider] must be set at detector stage
       this.notationProvider = this.column.temp[SeqTemps.notationProvider] ?? null;
@@ -185,7 +195,7 @@ export class SeqHandler implements ISeqHandler {
   public static setTags(uh: SeqHandler): void {
     const units = uh.column.meta.units as NOTATION;
 
-    if ([NOTATION.FASTA, NOTATION.SEPARATOR].includes(units)) {
+    if ([NOTATION.FASTA, NOTATION.SEPARATOR, NOTATION.BILN].includes(units)) {
       // Empty monomer alphabet is allowed, only if alphabet tag is annotated
       if (!uh.column.getTag(TAGS.alphabet) && Object.keys(uh.stats.freq).length === 0)
         throw new Error('Alphabet is empty and not annotated.');