@datagrok/sequence-translator 1.0.3 → 1.0.4

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@datagrok/sequence-translator",
   "friendlyName": "Sequence Translator",
-  "version": "1.0.3",
+  "version": "1.0.4",
   "description": "SequenceTranslator is a [package](https://datagrok.ai/help/develop/develop#packages) for the [Datagrok](https://datagrok.ai) platform, used to translate [oligonucleotide](https://en.wikipedia.org/wiki/Oligonucleotide) sequences between [different representations](https://github.com/datagrok-ai/public/tree/master/packages/SequenceTranslator#sequence-representations).",
   "repository": {
     "type": "git",
@@ -21,21 +21,21 @@
   },
   "scripts": {
     "link-api": "npm link datagrok-api",
-    "debug-sequencetranslator": "grok publish --rebuild",
-    "release-sequencetranslator": "grok publish --rebuild --release",
+    "debug-sequencetranslator": "grok publish",
+    "release-sequencetranslator": "grok publish localhost --release",
     "build-sequencetranslator": "webpack",
     "build": "webpack",
-    "debug-sequencetranslator-public": "grok publish public --rebuild",
-    "release-sequencetranslator-public": "grok publish public --rebuild --release",
-    "debug-sequencetranslator-local": "grok publish local --rebuild",
-    "release-sequencetranslator-local": "grok publish local --rebuild --release",
+    "debug-sequencetranslator-public": "grok publish public",
+    "release-sequencetranslator-public": "grok publish public --release",
+    "debug-sequencetranslator-local": "grok publish local",
+    "release-sequencetranslator-local": "grok publish local --release",
     "test": "set HOST=dev && jest",
     "test-dev": "set HOST=dev && jest",
     "test-local": "set HOST=localhost && jest"
   },
   "sources": [
     "css/style.css",
-    "common/openchemlib-full.js"
+    "vendors/openchemlib-full.js"
   ],
   "devDependencies": {
     "@typescript-eslint/eslint-plugin": "^4.29.1",

package/src/__jest__/remote.test.ts

@@ -16,34 +16,53 @@ beforeAll(async () => {
 }, P_START_TIMEOUT);
 
 afterAll(async () => {
-  await browser.close();
+  await browser?.close();
+});
+
+expect.extend({
+  checkOutput(received, expected, context) {
+    if (received === expected) {
+      return {
+        message: () => context,
+        pass: true
+      };
+    } else {
+      return {
+        message: () => context,
+        pass: false
+      };
+    }
+  }
 });
 
 it('TEST', async () => {
-  const target_package:string = process.env.TARGET_PACKAGE ?? 'SequenceTranslator';
-  console.log(`Testing ${target_package} package`);
+  const targetPackage:string = process.env.TARGET_PACKAGE ?? 'SequenceTranslator';
+  console.log(`Testing ${targetPackage} package`);
 
-  //console.log(require('root-require')('package.json').version);
-  let r = await page.evaluate((target_package):Promise<object> => {
+  let r = await page.evaluate((targetPackage):Promise<object> => {
     return new Promise<object>((resolve, reject) => {
-      (<any>window).grok.functions.eval(target_package + ':test()').then((df: any) => {
+      (<any>window).grok.functions.eval(targetPackage + ':test()').then((df: any) => {
         let cStatus = df.columns.byName('success');
         let cMessage = df.columns.byName('result');
         let cCat = df.columns.byName('category');
         let cName = df.columns.byName('name');
         let failed = false;
-        let report = '';
-        for (let i = 0; i < df.rowCount; i++)
-          if (!cStatus.get(i)) {
-            report += `${cCat.get(i)}.${cName.get(i)}: ${cMessage.get(i)}\n`;
+        let passReport = '';
+        let failReport = '';
+        for (let i = 0; i < df.rowCount; i++) {
+          if (cStatus.get(i)) {
+            passReport += `Test result : ${targetPackage}.${cCat.get(i)}.${cName.get(i)} : ${cMessage.get(i)}\n`;
+          } else {
             failed = true;
+            failReport += `Test result : ${targetPackage}.${cCat.get(i)}.${cName.get(i)} : ${cMessage.get(i)}\n`;
           }
-        resolve({report, failed});
+        }
+        resolve({failReport, passReport, failed});
       }).catch((e: any) => reject(e));
     });
-  }, target_package);
+  }, targetPackage);
   // @ts-ignore
-  console.log(r.report);
+  console.log(r.passReport);
   // @ts-ignore
-  expect(r.failed).toBe(false);
+  expect(r.failed).checkOutput(false, r.failReport);
 }, 100000);

package/src/defineAxolabsPattern.ts

@@ -82,7 +82,9 @@ function addColumnWithIds(tableName: string, columnName: string, patternName: st
   if (columns.contains(nameOfNewColumn))
     columns.remove(nameOfNewColumn);
   const columnWithIds = columns.byName(columnName);
-  return columns.addNewString(nameOfNewColumn).init((i: number) => columnWithIds.get(i) + '_' + patternName);
+  return columns.addNewString(nameOfNewColumn).init((i: number) => {
+    return (columnWithIds.getString(i) == '') ? '' : columnWithIds.get(i) + '_' + patternName;
+  });
 }
 
 function addColumnWithTranslatedSequences(
@@ -99,8 +101,10 @@ function addColumnWithTranslatedSequences(
     columns.remove(nameOfNewColumn);
   const columnWithInputSequences = columns.byName(columnName);
   return columns.addNewString(nameOfNewColumn).init((i: number) => {
-    return translateSequence(columnWithInputSequences.getString(i),
-      bases, ptoLinkages, startModification, endModification, firstPtoExist);
+    return columnWithInputSequences.getString(i) == '' ?
+      '' :
+      translateSequence(columnWithInputSequences.getString(i), bases, ptoLinkages, startModification, endModification,
+        firstPtoExist);
   });
 }
 
@@ -332,7 +336,7 @@ export function defineAxolabsPattern() {
     const col = tables.value!.columns.byName(colName);
     let allLengthsAreTheSame = true;
     for (let i = 1; i < col.length; i++) {
-      if (col.get(i - 1).length != col.get(i).length) {
+      if (col.get(i - 1).length != col.get(i).length && col.get(i).length != 0) {
         allLengthsAreTheSame = false;
         break;
       }
@@ -361,12 +365,13 @@ export function defineAxolabsPattern() {
   }
 
   async function postPatternToUserStorage() {
-    const author = await getCurrentUserName();
-    if (!saveAs.stringValue.includes('(created by '))
-      saveAs.value = saveAs.stringValue + author;
+    const currUserName = await getCurrentUserName();
+    saveAs.value = (saveAs.stringValue.includes('(created by ')) ?
+      getShortName(saveAs.value) + currUserName :
+      saveAs.stringValue + currUserName;
     return grok.dapi.userDataStorage.postValue(
       userStorageKey,
-      saveAs.stringValue,
+      saveAs.value,
       JSON.stringify({
         'ssBases': ssBases.slice(0, ssLength.value!).map((e) => e.value),
         'asBases': asBases.slice(0, asLength.value!).map((e) => e.value),
@@ -387,11 +392,12 @@ export function defineAxolabsPattern() {
       const lstMy: string[] = [];
       const lstOthers: string[] = [];
 
+      // TODO: display short name, but use long for querying userdataStorage
       for (const ent of Object.keys(entities)) {
         if (await isCurrentUserCreatedThisPattern(ent))
           lstOthers.push(ent);
         else
-          lstMy.push(getShortName(ent));
+          lstMy.push(ent);//getShortName(ent));
       }
 
       let loadPattern = ui.choiceInput('Load Pattern', '', lstMy, (v: string) => parsePatternAndUpdateUi(v));
@@ -526,7 +532,7 @@ export function defineAxolabsPattern() {
     const col = tables.value!.columns.byName(colName);
     if (col.type != DG.TYPE.INT)
       grok.shell.error('Column should contain integers only');
-    else if (col.categories.length < col.length) {
+    else if (col.categories.filter((e) => e != '').length < col.toList().filter((e) => e != '').length) {
       const duplicates = findDuplicates(col.getRawData());
       ui.dialog('Non-unique IDs')
         .add(ui.divText('Press \'OK\' to select rows with non-unique values'))
@@ -541,25 +547,43 @@ export function defineAxolabsPattern() {
   }
 
   const tables = ui.tableInput('Tables', grok.shell.tables[0], grok.shell.tables, (t: DG.DataFrame) => {
-    const inputSsColumn =
-      ui.choiceInput('SS Column', '', t.columns.names(), (colName: string) => validateSsColumn(colName));
+    const inputSsColumn = ui.choiceInput('SS Column', '', t.columns.names(), (colName: string) => {
+      validateSsColumn(colName);
+      ssVar = colName;
+    });
     inputSsColumnDiv.innerHTML = '';
     inputSsColumnDiv.append(inputSsColumn.root);
-    const inputAsColumn =
-      ui.choiceInput('AS Column', '', t.columns.names(), (colName: string) => validateAsColumn(colName));
+    const inputAsColumn = ui.choiceInput('AS Column', '', t.columns.names(), (colName: string) => {
+      validateAsColumn(colName);
+      asVar = colName;
+    });
     inputAsColumnDiv.innerHTML = '';
     inputAsColumnDiv.append(inputAsColumn.root);
-    const inputIdColumn =
-      ui.choiceInput('ID Column', '', t.columns.names(), (colName: string) => validateIdsColumn(colName));
+    const inputIdColumn = ui.choiceInput('ID Column', '', t.columns.names(), (colName: string) => {
+      validateIdsColumn(colName);
+      idVar = colName;
+    });
     inputIdColumnDiv.innerHTML = '';
     inputIdColumnDiv.append(inputIdColumn.root);
   });
 
-  const inputSsColumn = ui.choiceInput('SS Column', '', []);
+  let ssVar = '';
+  const inputSsColumn = ui.choiceInput('SS Column', '', [], (colName: string) => {
+    validateSsColumn(colName);
+    ssVar = colName;
+  });
   inputSsColumnDiv.append(inputSsColumn.root);
-  const inputAsColumn = ui.choiceInput('AS Column', '', []);
+  let asVar = '';
+  const inputAsColumn = ui.choiceInput('AS Column', '', [], (colName: string) => {
+    validateAsColumn(colName);
+    asVar = colName;
+  });
   inputAsColumnDiv.append(inputAsColumn.root);
-  const inputIdColumn = ui.choiceInput('ID Column', '', []);
+  let idVar = '';
+  const inputIdColumn = ui.choiceInput('ID Column', '', [], (colName: string) => {
+    validateIdsColumn(colName);
+    idVar = colName;
+  });
   inputIdColumnDiv.append(inputIdColumn.root);
 
   updatePatternsList();
@@ -638,7 +662,7 @@ export function defineAxolabsPattern() {
   });
 
   const convertSequenceButton = ui.button('Convert Sequences', () => {
-    if (inputSsColumn.value == null || (createAsStrand.value && inputAsColumn.value == null))
+    if (ssVar == '' || (createAsStrand.value && asVar == ''))
       grok.shell.info('Please select table and columns on which to apply pattern');
     else if (ssLength.value != ssInputExample.value.length || asLength.value != asInputExample.value.length) {
       const dialog = ui.dialog('Length Mismatch');
@@ -652,14 +676,14 @@ export function defineAxolabsPattern() {
         })
         .show();
     } else {
-      if (inputIdColumn.value != null)
-        addColumnWithIds(tables.value!.name, inputIdColumn.value, getShortName(saveAs.value));
+      if (idVar != '')
+        addColumnWithIds(tables.value!.name, idVar, getShortName(saveAs.value));
       addColumnWithTranslatedSequences(
-        tables.value!.name, inputSsColumn.value, ssBases, ssPtoLinkages,
+        tables.value!.name, ssVar, ssBases, ssPtoLinkages,
         ssFiveModification, ssThreeModification, firstSsPto.value!);
       if (createAsStrand.value) {
         addColumnWithTranslatedSequences(
-          tables.value!.name, inputAsColumn.value!, asBases, asPtoLinkages,
+          tables.value!.name, asVar, asBases, asPtoLinkages,
           asFiveModification, asThreeModification, firstAsPto.value!);
       }
       grok.shell.v = grok.shell.getTableView(tables.value!.name);

package/src/package.ts

@@ -2,7 +2,6 @@
 import * as grok from 'datagrok-api/grok';
 import * as ui from 'datagrok-api/ui';
 import * as DG from 'datagrok-api/dg';
-import * as OCL from 'openchemlib/full.js';
 import $ from 'cash-dom';
 import {defineAxolabsPattern} from './defineAxolabsPattern';
 import {saveSenseAntiSense} from './structures-works/save-sense-antisense';
@@ -10,7 +9,8 @@ import {sequenceToSmiles, sequenceToMolV3000} from './structures-works/from-mono
 import {convertSequence, undefinedInputSequence, isValidSequence, getFormat} from
   './structures-works/sequence-codes-tools';
 import {map, COL_NAMES, MODIFICATIONS} from './structures-works/map';
-import {siRnaAxolabsToGcrs} from './structures-works/converters';
+import {siRnaAxolabsToGcrs, gcrsToNucleotides, asoGapmersBioSpringToGcrs, gcrsToMermade12,
+} from './structures-works/converters';
 import {SALTS_CSV} from './salts';
 import {USERS_CSV} from './users';
 import {ICDS} from './ICDs';
@@ -309,23 +309,43 @@ function molecularWeight(sequence: string, weightsObj: {[index: string]: number}
 
 //tags: autostart
 export function autostartOligoSdFileSubscription() {
-  let alreadyAdded = false;
   grok.events.onViewAdded.subscribe((v: any) => {
     if (v.type == 'TableView') {
       if (v.dataFrame.columns.contains(COL_NAMES.TYPE))
         oligoSdFile(v.dataFrame);
       grok.events.onContextMenu.subscribe((args) => {
-        for (const col of v.dataFrame.columns) {
-          if (!alreadyAdded && DG.Detector.sampleCategories(col, (s) => /^[fsACGUacgu]{6,}$/.test(s))) {
-            alreadyAdded = true;
-            args.args.menu.item('Convert to GCRS', () => {
-              const seqCol = args.args.context.table.currentCol;
-              args.args.context.table.columns.addNewString(seqCol.name + ' to GCRS').init((i: number) => {
-                return siRnaAxolabsToGcrs(seqCol.get(i));
-              });
+        const seqCol = args.args.context.table.currentCol;
+        if (DG.Detector.sampleCategories(seqCol, (s) => /^[fsACGUacgu]{6,}$/.test(s))) {
+          args.args.menu.item('Convert Axolabs to GCRS', () => {
+            args.args.context.table.columns.addNewString(seqCol.name + ' to GCRS').init((i: number) => {
+              return siRnaAxolabsToGcrs(seqCol.get(i));
             });
-            break;
-          };
+          });
+        } else if (DG.Detector.sampleCategories(seqCol, (s) => /^[fmpsACGU]{6,}$/.test(s)) ||
+        DG.Detector.sampleCategories(seqCol, (s) => /^(?=.*moe)(?=.*5mC)(?=.*ps){6,}/.test(s))) {
+          args.args.menu.item('Convert GCRS to raw', () => {
+            args.args.context.table.columns.addNewString(seqCol.name + ' to raw').init((i: number) => {
+              return gcrsToNucleotides(seqCol.get(i));
+            });
+          });
+          args.args.menu.item('Convert GCRS to MM12', () => {
+            args.args.context.table.columns.addNewString(seqCol.name + ' to MM12').init((i: number) => {
+              return gcrsToMermade12(seqCol.get(i));
+            });
+          });
+        } else if (DG.Detector.sampleCategories(seqCol, (s) => /^[*56789ATGC]{6,}$/.test(s))) {
+          args.args.menu.item('Convert Biospring to GCRS', () => {
+            const seqCol = args.args.context.table.currentCol;
+            args.args.context.table.columns.addNewString(seqCol.name + ' to GCRS').init((i: number) => {
+              return asoGapmersBioSpringToGcrs(seqCol.get(i));
+            });
+          });
+        } else if (DG.Detector.sampleCategories(seqCol, (s) => /^[*1-8]{6,}$/.test(s))) {
+          args.args.menu.item('Convert Biospring to GCRS', () => {
+            args.args.context.table.columns.addNewString(seqCol.name + ' to GCRS').init((i: number) => {
+              return siRnaAxolabsToGcrs(seqCol.get(i));
+            });
+          });
         }
       });
     }

package/src/structures-works/mol-transformations.ts

@@ -1,5 +1,3 @@
-import * as OCL from 'openchemlib/full.js';
-
 const PHOSHATE = `
 Datagrok monomer library Nucleotides
 
@@ -254,6 +252,7 @@ export function linkV3000(molBlocks: string[], twoChains: boolean = false, oclRe
 }
 
 function rotateNucleotidesV3000(molecule: string) {
+  // @ts-ignore
   let molBlock = molecule.includes('M  END') ? molecule : OCL.Molecule.fromSmiles(molecule).toMolfileV3();
   const coordinates = extractAtomDataV3000(molBlock);
   const natom = coordinates.atomIndex.length;
@@ -320,6 +319,7 @@ function rotateNucleotidesV3000(molecule: string) {
 }
 
 function invertNucleotidesV3000(molecule: string) {
+  // @ts-ignore
   let molBlock = molecule.includes('M  END') ? molecule : OCL.Molecule.fromSmiles(molecule).toMolfileV3();
   const coordinates = extractAtomDataV3000(molBlock);
   const natom = coordinates.atomIndex.length;