@datagrok-libraries/utils 0.0.5 → 0.0.6

package/package.json

@@ -4,7 +4,7 @@
 		"access": "public"
 	},
 	"fullName": "Utils",
-	"version": "0.0.5",
+	"version": "0.0.6",
 	"description": "Common utilities",
 	"dependencies": {
 		"@keckelt/tsne": "^1.0.2",
@@ -13,9 +13,10 @@
 		"dts-bundle": "^0.7.3",
 		"fastest-levenshtein": "^1.0.12",
 		"jaro-winkler-typescript": "^1.0.1",
-		"typescript": "^3.9.7",
+		"typescript": "^4.2.3",
 		"umap-js": "^1.3.3",
-		"js-yaml": "^4.1.0"
+		"js-yaml": "^4.1.0",
+		"node-fetch": "^2.6.0"
 	},
 	"scripts": {
 		"link": "npm link",

package/src/bit-array.ts

@@ -53,28 +53,39 @@ export default class BitArray {
     7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7,
     7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7]);
 
-  _data: Uint32Array;
-  _length = 0;
-  _version = 0;
-  _updateLevel = 0;
-  _selectedCount = 0;
-  _selectedCountVersion = -1;
-  _selectedIndexesVersion = -1;
-  _versionedName = '';
-  _versionedNameVersion = -1;
+  private _data: Uint32Array;
+  private _length = 0;
+  private _version = 0;
+  private _updateLevel = 0;
+  private _selectedCount = 0;
+  private _selectedCountVersion = -1;
+  private _selectedIndexesVersion = -1;
+  private _versionedName = '';
+  private _versionedNameVersion = -1;
   SHRINK_THRESHOLD = 0x100;
 
-  constructor(length: number, defaultValue = false) {
-    this._length = length;
-    this._data = BitArray._createBuffer(length);
-
-    if (defaultValue) {
-      for (let i = 0; i < this._data.length; i++) {
-        this._data[i] = -1;
-      }
+  constructor(data: Uint32Array, length: number)
+  constructor(length: number, defaultValue?: boolean)
+  constructor(arg: number | Uint32Array, defaultValue: boolean | number = false) {
+    if (typeof arg === 'number') {
+      const length = arg;
+      let buff = BitArray._createBuffer(length);
+      if (defaultValue)
+        for (let i = 0; i < buff.length; i++)
+          buff[i] = -1;
+      this._data = buff;
+      this._length = length;
+    } else if (arg instanceof Uint32Array) {
+      this._data = arg as Uint32Array;
+      this._length = defaultValue as number;
+    } else {
+      throw new Error("Invalid constructor");
     }
   }
 
+
+  getRawData() { return this._data; }
+
   assureGoez(num: number, argName: String): void {
     if (num < 0) throw new Error(`${argName} should be greater than zero`);
   }
@@ -181,7 +192,7 @@ export default class BitArray {
     return temp;
   }
 
-  static _createBuffer(length: number): Uint32Array {
+  private static _createBuffer(length: number): Uint32Array {
     return new Uint32Array(Math.floor((length + 0x1f) / 0x20));
   }
 

package/src/reduce-dimensionality.ts

@@ -5,7 +5,8 @@ import {Options, DistanceMetric, Coordinates, Vector, Vectors, Matrix} from './t
 import {calcDistanceMatrix, transposeMatrix} from './operations';
 import {SPEBase, PSPEBase} from './spe';
 import {Measurer} from './string-measure';
-//import {AlignedSequenceEncoder} from './sequence-encoder';
+import {AlignedSequenceEncoder} from './sequence-encoder';
+import {assert} from './operations';
 
 /**
  * Abstract dimensionality reducer.
@@ -80,8 +81,9 @@ class TSNEReducer extends Reducer {
  */
 class UMAPReducer extends Reducer {
   protected reducer: umj.UMAP;
-/*  protected encoder: AlignedSequenceEncoder;
-  protected distanceMatrix: Matrix;*/
+  protected encoder: AlignedSequenceEncoder;
+  protected distanceFn: Function;
+  protected vectors: number[][];
 
   /**
    * Creates an instance of UMAPReducer.
@@ -90,46 +92,54 @@ class UMAPReducer extends Reducer {
    */
   constructor(options: Options) {
     super(options);
-    this.reducer = new umj.UMAP(options);
-/*    this.encoder = new AlignedSequenceEncoder();
-    this.distanceMatrix = calcDistanceMatrix(this.data, options.distanceFn);
-    this.distanceFn = options.distanceFn;*/
-  }
 
-/*  protected _encodedDistance(a: number[], b: number[]): number {
-    return this.distanceMatrix[a[a.length-1]][b[b.length-1]];
-  }
+    assert('distanceFn' in options);
 
-  protected _decodedDistance(a: number[], b: number[]): number {
-    return this.encoder.decode(a);
+    this.encoder = new AlignedSequenceEncoder();
+    this.distanceFn = options.distanceFn;
+    this.vectors = [];
+    options.distanceFn = this._encodedDistance.bind(this);
+    this.reducer = new umj.UMAP(options);
   }
 
-  protected encode(): number[][] {
-    let vectors: number[][] = [];
+  /**
+   * Custom distance wrapper to have numeric inputs instead of string ones.
+   *
+   * @protected
+   * @param {number[]} a The first item.
+   * @param {number[]} b The first item.
+   * @return {number} Distance metric.
+   * @memberof UMAPReducer
+   */
+  protected _encodedDistance(a: number[], b: number[]): number {
+    return this.distanceFn(this.data[a[0]], this.data[b[0]]);
+  }
 
+  /**
+   * Encodes the input data as a vector of indices.
+   *
+   * @protected
+   * @memberof UMAPReducer
+   */
+  protected _encode() {
     for (let i = 0; i < this.data.length; ++i) {
-      let encoded = this.encoder.encode(this.data[i]);
-      vectors.push(encoded);
+      this.vectors.push([i]);
     }
-    return vectors;
   }
 
-  protected _calcDistance() {
-    const nItems = this.data.length;
-    const vectors = this.encode();
-    
-  }*/
-
   /**
    * Embeds the data given into the two-dimensional space using UMAP method.
    * @return {Coordinates} Cartesian coordinate of this embedding.
    */
   public transform(): Coordinates {
-    const embedding = this.reducer.fit(this.data);
+    this._encode();
+
+    const embedding = this.reducer.fit(this.vectors);
 
     function arrayCast2Coordinates(data: number[][]): Coordinates {
       return new Array(data.length).fill(0).map((_, i) => (Vector.from(data[i])));
     }
+
     return arrayCast2Coordinates(embedding);
   }
 }
@@ -198,7 +208,7 @@ class PSPEReducer extends Reducer {
  */
 export class DimensionalityReducer {
   private reducer: Reducer | undefined;
-  private methods: string[];
+  public static methods: string[] = ['UMAP', 't-SNE', 'SPE', 'pSPE'];
   private measurer: Measurer;
 
   /**
@@ -210,11 +220,7 @@ export class DimensionalityReducer {
    * @memberof DimensionalityReducer
    */
   constructor(data: any[], method: string, metric: string, options?: Options) {
-    this.methods = ['UMAP', 'TSNE', 'SPE', 'PSPE'];
-
-    if (!this.availableMethods.includes(method)) {
-      throw new Error('The method "'+method+'" is not supported');
-    }
+    assert(DimensionalityReducer.availableMethods.includes(method), `The method '${method}' is not supported`);
 
     this.measurer = new Measurer(metric);
     const measure = this.measurer.getMeasure()
@@ -226,7 +232,7 @@ export class DimensionalityReducer {
         ...{nEpochs: options?.cycles},
         ...options
       });
-    } else if (method == 'TSNE') {
+    } else if (method == 't-SNE') {
       this.reducer = new TSNEReducer({
         ...{data: data},
         ...{distance: measure},
@@ -266,7 +272,7 @@ export class DimensionalityReducer {
    * @readonly
    * @memberof DimensionalityReducer
    */
-  get availableMethods() {
-    return this.methods;
+  static get availableMethods() {
+    return DimensionalityReducer.methods;
   }
 }

package/src/sequence-encoder.ts

@@ -41,12 +41,12 @@ export class AlignedSequenceEncoder {
     /**
      * Truncate NH2 and -COOH terminals of the given sequence.
      *
-     * @protected
+     * @static
      * @param {string} seq The sequence provided.
      * @return {string} Truncated sequence.
      * @memberof AlignedSequenceEncoder
      */
-    protected _truncateSequence(seq: string): string {
+    static _truncateSequence(seq: string): string {
         let start = 0;
         let end = seq.length;
         const termina = ['NH2', 'COOH'];
@@ -67,24 +67,25 @@ export class AlignedSequenceEncoder {
     /**
      * Cuts auxiliary defises before a residue.
      *
-     * @protected
+     * @static
      * @param {string} seq The sequence to process.
      * @return {string} Processed sequence.
      * @memberof AlignedSequenceEncoder
      */
-    protected _dropDefises(seq: string): string {
+    static _dropDefises(seq: string): string {
         return seq.replace(/(-)([^-]+)/g, '$2')
     }
 
     /**
      * Performs truncation and cutting auxiliary defises.
      *
+     * @static
      * @param {string} sequence The sequence work under process.
      * @return {string} Result of cleaning.
      * @memberof AlignedSequenceEncoder
      */
-    public clean(sequence: string): string {
-        return this._dropDefises(this._truncateSequence(sequence));
+    static clean(sequence: string): string {
+        return AlignedSequenceEncoder._dropDefises(AlignedSequenceEncoder._truncateSequence(sequence));
     }
 
     /**
@@ -95,7 +96,7 @@ export class AlignedSequenceEncoder {
      * @memberof AlignedSequenceEncoder
      */
     public encode(sequence: string): number[] {
-        const seq = this.clean(sequence);
+        const seq = AlignedSequenceEncoder.clean(sequence);
         const nItems = seq.length;
         let values = new Array(nItems).fill(0);
     

package/src/string-measure.ts

@@ -12,7 +12,7 @@ import {assert} from './operations';
  */
 export class Measurer {
     protected method: string;
-    protected receipt: {[name: string]: DistanceMetric} = {
+    public static receipt: {[name: string]: DistanceMetric} = {
       'Levenshtein': fl.distance,
       'Jaro-Winkler': jaroWinkler,
     };
@@ -23,7 +23,7 @@ export class Measurer {
      * @memberof Measurer
      */
     constructor(method: string) {
-      assert(this.availableMeasures.includes(method), 'The ${method} was not found.')
+      assert(Measurer.availableMeasures.includes(method), 'The ${method} was not found.')
       this.method = method;
     }
 
@@ -32,7 +32,7 @@ export class Measurer {
      * @return {DistanceMetric} Callback of the measure chosen.
      */
     public getMeasure(): DistanceMetric {
-      return this.receipt[this.method];
+      return Measurer.receipt[this.method];
     }
 
     /**
@@ -42,10 +42,10 @@ export class Measurer {
      * @type {string[]}
      * @memberof Measurer
      */
-    public get availableMeasures() : string[] {
+    public static get availableMeasures() : string[] {
       let keys = [];
 
-      for (let key in this.receipt) {
+      for (let key in Measurer.receipt) {
         keys.push(key);
       }
       return keys;