@datagrok/bio 2.10.2 → 2.10.4

package/link-bio

@@ -0,0 +1,7 @@
+#!/bin/bash
+
+cd ../../libraries/bio
+./cleanup && npm i && npm run build && npm link
+
+cd -
+npm link @datagrok-libraries/bio

package/package.json

@@ -1,11 +1,12 @@
 {
   "name": "@datagrok/bio",
   "friendlyName": "Bio",
+  "skipCI": true,
   "author": {
     "name": "Leonid Stolbov",
     "email": "lstolbov@datagrok.ai"
   },
-  "version": "2.10.2",
+  "version": "2.10.4",
   "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",
@@ -34,7 +35,7 @@
   ],
   "dependencies": {
     "@biowasm/aioli": "^3.1.0",
-    "@datagrok-libraries/bio": "^5.38.2",
+    "@datagrok-libraries/bio": "^5.38.3",
     "@datagrok-libraries/chem-meta": "^1.0.1",
     "@datagrok-libraries/ml": "^6.3.39",
     "@datagrok-libraries/tutorials": "^1.3.6",

package/setup

@@ -0,0 +1,52 @@
+#!/bin/bash
+
+log() {
+  GREEN='\e[0;32m'
+  NO_COLOR='\e[0m'
+  echo -e $GREEN $1 $NO_COLOR
+}
+
+unlink_cleanup() {
+  npm uninstall --location=global datagrok-api @datagrok-libraries/utils @datagrok-libraries/ml @datagrok-libraries/bio
+
+  for dir in ${dirs[@]}; do
+    echo "Removing node_modules and dist in $(pwd)"
+    cd $package_dir && cd $dir
+    rm -rf node_modules dist
+  done
+}
+
+package_dir=$(pwd)
+
+dirs=(
+  "../../js-api/"
+  "../../libraries/utils/"
+  "../../libraries/ml/"
+  "../../libraries/bio/"
+)
+
+unlink_cleanup
+
+for dir in ${dirs[@]}; do
+  cd $package_dir && cd $dir
+  if [ $dir == "../../js-api/" ]; then
+   git clean -f -X -d ./src
+  fi
+  log "npm install in $(pwd)"
+  npm install
+  log "npm link in $(pwd)"
+  npm link
+done
+
+for dir in ${dirs[@]}; do
+  cd $package_dir && cd $dir
+  if [ $dir != "../../js-api/" ]; then
+    log "npm link-all in $(pwd)"
+    npm run link-all
+  fi
+  log "npm run build in $(pwd)"
+  npm run build || exit
+done
+
+cd $package_dir
+npm run link-all

package/src/package.ts

@@ -65,6 +65,7 @@ import {PackageSettingsEditorWidget} from './widgets/package-settings-editor-wid
 import {getCompositionAnalysisWidget} from './widgets/composition-analysis-widget';
 import {MacromoleculeColumnWidget} from './utils/macromolecule-column-widget';
 import {addCopyMenuUI} from './utils/context-menu';
+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';
@@ -97,10 +98,9 @@ export class SeqPaletteCustom implements SeqPalette {
 
 //tags: init
 export async function initBio() {
-  let module: RDModule;
+  const module = await grok.functions.call('Chem:getRdKitModule');
   await Promise.all([
     (async () => { await MonomerLibHelper.instance.loadLibraries(); })(),
-    (async () => { module = await grok.functions.call('Chem:getRdKitModule'); })(),
     (async () => {
       const pkgProps = await _package.getProperties();
       const bioPkgProps = new BioPackageProperties(pkgProps);
@@ -981,3 +981,19 @@ export async function demoBioAtomicLevel(): Promise<void> {
 export async function demoBioHelmMsaSequenceSpace(): Promise<void> {
   await demoBio05UI();
 }
+
+//name: enumeratorColumnChoice
+//input: dataframe df [Input data table]
+//input: column macroMolecule
+export async function enumeratorColumnChoice(df: DG.DataFrame, macroMolecule: DG.Column): Promise<void> {
+  _setPeptideColumn(macroMolecule);
+  await grok.data.detectSemanticTypes(df);
+}
+
+//name: PolyTool
+//input: column molColumn {semType: Macromolecule}
+//tags: panel, exclude-actions-panel
+//output: widget result
+export function getEnumeratorWidget(molColumn: DG.Column): DG.Widget {
+  return _getEnumeratorWidget(molColumn);
+}

package/src/utils/enumerator-tools.ts

@@ -0,0 +1,162 @@
+
+import * as grok from 'datagrok-api/grok';
+import * as ui from 'datagrok-api/ui';
+import * as DG from 'datagrok-api/dg';
+
+import {NOTATION, ALIGNMENT, ALPHABET} from '@datagrok-libraries/bio/src/utils/macromolecule';
+import {UnitsHandler} from '@datagrok-libraries/bio/src/utils/units-handler';
+import {HELM_POLYMER_TYPE} from '@datagrok-libraries/bio/src/utils/const';
+import {MonomerLibHelper} from '../utils/monomer-lib';
+import {_package} from '../package';
+import * as rxjs from 'rxjs';
+
+const LEFT_HELM_WRAPPER = 'PEPTIDE1{';
+const RIGHT_HELM_WRAPPER = '}$$$$';
+const ALL_MONOMERS = '<All>';
+
+const enum CYCLIZATION_TYPE {
+  NO = 'N-O',
+  R3 = 'R3-R3',
+}
+
+function addCommonTags(col: DG.Column):void {
+  col.setTag('quality', DG.SEMTYPE.MACROMOLECULE);
+  col.setTag('aligned', ALIGNMENT.SEQ);
+  col.setTag('alphabet', ALPHABET.PT);
+}
+
+export function _setPeptideColumn(col: DG.Column): void {
+  addCommonTags(col);
+  col.setTag('units', NOTATION.SEPARATOR);
+  col.setTag('separator', '-');
+  // col.setTag('cell.renderer', 'sequence');
+}
+
+async function enumerator(
+  molColumn: DG.Column, cyclizationType: CYCLIZATION_TYPE, leftTerminal: string, rightTerminal: string
+): Promise<void> {
+  function hasR3Terminals(helm: string, leftTerminal: string, rightTerminal: string): boolean {
+    if (leftTerminal === ALL_MONOMERS || rightTerminal === ALL_MONOMERS)
+      return true;
+    const positions = getLinkedR3Positions(helm);
+    return positions.every((el) => el > 0);
+  }
+
+  function hasNOTerminals(helm: string, leftTerminal: string, rightTerminal: string): boolean {
+    if (leftTerminal === ALL_MONOMERS || rightTerminal === ALL_MONOMERS)
+      return true;
+    return helm.includes(LEFT_HELM_WRAPPER + leftTerminal) && helm.includes(rightTerminal + RIGHT_HELM_WRAPPER);
+  }
+
+  function applyModification(helm: string): string {
+    if (cyclizationType === CYCLIZATION_TYPE.R3)
+      return applyR3Modification(helm);
+    else
+      return applyNOModification(helm);
+  }
+
+  function applyNOModification(helm: string): string {
+    if (hasNOTerminals(helm, leftTerminal, rightTerminal))
+      return getNOCycle(helm, getLinkedNOPositions(helm));
+    return helm;
+  }
+
+  function applyR3Modification(helm: string): string {
+    if (hasR3Terminals(helm, leftTerminal, rightTerminal))
+      return getR3Cycle(helm, getLinkedR3Positions(helm));
+    return helm;
+  }
+
+  function getLinkedR3Positions(helm: string): [number, number] {
+    const seq = helm.replace(LEFT_HELM_WRAPPER, '').replace(RIGHT_HELM_WRAPPER, '');
+    const monomers = seq.split('.');
+    const start = monomers.findIndex((el) => el === leftTerminal);
+    const end = monomers.findIndex((el, idx) => el === rightTerminal && idx > start);
+    return [start + 1, end + 1];
+  }
+
+  function getLinkedNOPositions(helm: string): [number, number] {
+    const seq = helm.replace(LEFT_HELM_WRAPPER, '').replace(RIGHT_HELM_WRAPPER, '');
+    const lastMonomerNumber = seq.split('.').length;
+    return [1, lastMonomerNumber];
+  }
+
+  function getR3Cycle(helm: string, position: [number, number]): string {
+    const result = helm.replace(RIGHT_HELM_WRAPPER,
+      `}$PEPTIDE1,PEPTIDE1,${position[0]}:R3-${position[1]}:R3${'$'.repeat(6)}`);
+    return result;
+  }
+
+  function getNOCycle(helm: string, position: [number, number]): string {
+    const result = helm.replace(RIGHT_HELM_WRAPPER,
+      `}$PEPTIDE1,PEPTIDE1,${position[1]}:R2-${position[0]}:R1${'$'.repeat(6)}`);
+    return result;
+  }
+
+  const df = molColumn.dataFrame;
+  const uh = UnitsHandler.getOrCreate(molColumn);
+  const sourceHelmCol = uh.convert(NOTATION.HELM);
+  const targetList = sourceHelmCol.toList().map((helm) => applyModification(helm));
+  const colName = df.columns.getUnusedName('Cyclization(' + molColumn.name + ')');
+  const targetHelmCol = DG.Column.fromList('string', colName, targetList);
+
+  addCommonTags(targetHelmCol);
+  targetHelmCol.setTag('units', NOTATION.HELM);
+  targetHelmCol.setTag('cell.renderer', 'helm');
+
+  df.columns.add(targetHelmCol);
+  await grok.data.detectSemanticTypes(df);
+}
+
+export function _getEnumeratorWidget(molColumn: DG.Column): DG.Widget {
+  function updateMonomerList(): void {
+    console.log('hi from update:');
+    if (cyclizationTypeChoice.value === cyclizationTypes[0]) {
+      console.log('hi from first branch:');
+      monomerList = [ALL_MONOMERS].concat(
+        monomerLib.getMonomerSymbolsByType(HELM_POLYMER_TYPE.PEPTIDE)
+      );
+    } else if (cyclizationTypeChoice.value === cyclizationTypes[1]) {
+      monomerList = [ALL_MONOMERS].concat(
+        monomerLib.getMonomerSymbolsByRGroup(3, HELM_POLYMER_TYPE.PEPTIDE)
+      );
+      console.log('hi from second branch:');
+    }
+    leftTerminalChoice = ui.choiceInput('R1:', monomerList[0], monomerList);
+    rightTerminalChoice = ui.choiceInput('R2:', monomerList[0], monomerList);
+    ui.empty(terminalControls);
+    [leftTerminalChoice, rightTerminalChoice].forEach((el) => { terminalControls.appendChild(el.root); });
+  }
+
+  const onCyclizationChoice = new rxjs.Subject<string>();
+  onCyclizationChoice.subscribe(() => updateMonomerList());
+
+  const modifications = ['Cyclization'];
+  const modificationChoice = ui.choiceInput('Modification', modifications[0], modifications);
+
+  const cyclizationTypes = [CYCLIZATION_TYPE.NO, CYCLIZATION_TYPE.R3];
+  const cyclizationTypeChoice = ui.choiceInput(
+    'Type', cyclizationTypes[0], cyclizationTypes, () => { onCyclizationChoice.next(); }
+  );
+
+  const monomerLib = MonomerLibHelper.instance.getBioLib();
+  let monomerList: string[] = [];
+  let leftTerminalChoice = ui.choiceInput('R1:', monomerList[0], monomerList);
+  let rightTerminalChoice = ui.choiceInput('R2:', monomerList[0], monomerList);
+  const terminalControls = ui.divV([leftTerminalChoice.root, rightTerminalChoice.root]);
+
+  updateMonomerList();
+
+  const btn = ui.bigButton('Run', async () =>
+    enumerator(molColumn, cyclizationTypeChoice.value!, leftTerminalChoice.value!, rightTerminalChoice.value!)
+  );
+
+  const div = ui.div([
+    modificationChoice,
+    cyclizationTypeChoice,
+    terminalControls,
+    btn
+  ]);
+
+  return new DG.Widget(div);
+}

package/src/utils/monomer-lib.ts

@@ -5,11 +5,16 @@ import * as DG from 'datagrok-api/dg';
 import {Observable, Subject} from 'rxjs';
 
 import {IMonomerLib, Monomer} from '@datagrok-libraries/bio/src/types/index';
+import {MolfileHandler} from '@datagrok-libraries/chem-meta/src/parsing-utils/molfile-handler';
 import {
   createJsonMonomerLibFromSdf,
+  getJsonMonomerLibForEnumerator,
   IMonomerLibHelper,
+  isValidEnumeratorLib,
 } from '@datagrok-libraries/bio/src/monomer-works/monomer-utils';
-import {HELM_REQUIRED_FIELDS as REQ, HELM_OPTIONAL_FIELDS as OPT} from '@datagrok-libraries/bio/src/utils/const';
+import {
+  HELM_REQUIRED_FIELDS as REQ, HELM_OPTIONAL_FIELDS as OPT, HELM_POLYMER_TYPE
+} from '@datagrok-libraries/bio/src/utils/const';
 
 import {_package} from '../package';
 
@@ -135,6 +140,39 @@ export class MonomerLib implements IMonomerLib {
     return Object.keys(this._monomers[polymerType]);
   }
 
+  /** Get a list of monomers with specified element attached to specified
+   * R-group
+   * WARNING: RGroup numbering starts from 1, not 0*/
+  getMonomerSymbolsByRGroup(rGroupNumber: number, polymerType: string, element?: string): string[] {
+    const monomerSymbols = this.getMonomerSymbolsByType(polymerType);
+    let monomers = monomerSymbols.map((sym) => this.getMonomer(polymerType, sym));
+    monomers = monomers.filter((el) => el !== null);
+    if (monomers.length === 0)
+      return [];
+
+    function findAllIndices<T>(arr: T[], element: T): number[] {
+      return arr.map((value, index) => (value === element ? index : -1))
+        .filter((index) => index !== -1);
+    }
+
+    monomers = monomers.filter((monomer) => {
+      if (!monomer?.rgroups)
+        return false;
+      console.log('monomer symbol:', monomer.symbol);
+      let criterion = monomer?.rgroups.length >= rGroupNumber;
+      console.log(`has rGroupNumber ${rGroupNumber}`, criterion);
+      const molfileHandler = MolfileHandler.getInstance(monomer.molfile);
+      console.log(molfileHandler.atomTypes);
+      const rGroupIndices = findAllIndices(molfileHandler.atomTypes, 'R#');
+      console.log('rGroup indices', rGroupIndices);
+      console.log(molfileHandler.pairsOfBondedAtoms);
+      criterion &&= true;
+      console.log('criterion', criterion);
+      return criterion;
+    });
+    return monomers.map((monomer) => monomer?.symbol!);
+  }
+
   get onChanged(): Observable<any> {
     return this._onChanged;
   }
@@ -244,9 +282,27 @@ export class MonomerLibHelper implements IMonomerLibHelper {
       } else {
         grok.shell.warning('Chem package is not installed');
       }
-    } else {
+    } else if (fileName.endsWith('.json')) {
       const file = await fileSource.readAsText(fileName);
       rawLibData = JSON.parse(file);
+    } else if (fileName.endsWith('.csv')) {
+      // todo: replace by DataFrame's method after update of js-api
+      function toJson(df: DG.DataFrame): any[] {
+        return Array.from({length: df.rowCount}, (_, idx) =>
+          df.columns.names().reduce((entry: {[key: string]: any}, colName) => {
+            entry[colName] = df.get(colName, idx);
+            return entry;
+          }, {})
+        );
+      }
+      const df = await fileSource.readCsv(fileName);
+      const json = toJson(df);
+      if (isValidEnumeratorLib(json))
+        rawLibData = getJsonMonomerLibForEnumerator(json);
+      else
+        throw new Error('Invalid format of CSV monomer lib');
+    } else {
+      throw new Error('Monomer library of unknown file format, supported formats: SDF, JSON, CSV');
     }
 
     const monomers: { [polymerType: string]: { [monomerSymbol: string]: Monomer } } = {};

package/setup-unlink-clean.sh

@@ -1,22 +0,0 @@
-#!/bin/bash
-package_dir=$(pwd)
-
-GREEN='\e[0;32m'
-NO_COLOR='\e[0m'
-
-dirs=(
-  "../../js-api/"
-  "../../libraries/utils/"
-  "../../libraries/ml/"
-  "../../libraries/bio/"
-)
-
-npm uninstall --location=global datagrok-api @datagrok-libraries/utils @datagrok-libraries/ml @datagrok-libraries/bio
-
-for dir in ${dirs[@]}; do
-  cd $package_dir
-  cd $dir 
-  echo -e $GREEN Removing node_modules and dist in $(pwd) $NO_COLOR
-  rm -rf node_modules dist
-  # rm package-lock.json
-done

package/setup.sh

@@ -1,38 +0,0 @@
-#!/bin/bash
-
-./setup-unlink-clean.sh
-
-GREEN='\e[0;32m'
-NO_COLOR='\e[0m'
-
-package_dir=$(pwd)
-
-dirs=(
-  "../../js-api/"
-  "../../libraries/utils/"
-  "../../libraries/ml/"
-  "../../libraries/bio/"
-)
-
-for dir in ${dirs[@]}; do
-  cd $package_dir
-  cd $dir 
-  echo -e $GREEN npm install in $(pwd) $NO_COLOR
-  npm install 
-  echo -e $GREEN npm link in $(pwd) $NO_COLOR
-  npm link
-done
-
-for dir in ${dirs[@]}; do
-  cd $package_dir
-  cd $dir 
-  if [ $dir != "../../js-api/" ]; then
-    echo -e $GREEN npm link-all in $(pwd) $NO_COLOR
-    npm run link-all
-  fi
-  echo -e $GREEN npm run build in$(pwd) $NO_COLOR
-  npm run build || exit
-done
-
-cd $package_dir
-npm run link-all