VALIS-40-update-genome-preprocess

tools/genome-preprocess/dist/tools/genome-preprocess/src/FileSystemUtils.js

@@ -1,5 +1,6 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
+exports.deleteDirectory = void 0;
 const Terminal_1 = require("./Terminal");
 const fs = require("fs");
 function deleteDirectory(directory) {

tools/genome-preprocess/dist/tools/genome-preprocess/src/Terminal.js

@@ -1,5 +1,6 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
+exports.Terminal = void 0;
 const util = require("util");
 const process = require("process");
 class Terminal {
@@ -56,8 +57,8 @@ class Terminal {
         }
     }
 }
-Terminal.currentRewriteId = undefined;
 exports.Terminal = Terminal;
+Terminal.currentRewriteId = undefined;
 exports.default = Terminal;
 var FormatFlag;
 (function (FormatFlag) {

tools/genome-preprocess/dist/tools/genome-preprocess/src/Tileset.js

@@ -1,5 +1,6 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
+exports.Tileset = void 0;
 class Tileset {
     constructor(tileSize) {
         this.tileSize = tileSize;

tools/genome-preprocess/dist/tools/genome-preprocess/src/gff3/AnnotationTileset.js

@@ -1,6 +1,7 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-const AnnotationTypes_1 = require("../../../../src/track/annotation/AnnotationTypes");
+exports.AnnotationTileset = void 0;
+const AnnotationTypes_1 = require("./AnnotationTypes");
 class AnnotationTileset {
     constructor(tileSize, topLevelOnly, onUnknownFeature, onError) {
         this.tileSize = tileSize;
@@ -40,17 +41,17 @@ class AnnotationTileset {
                 let isTranscript = AnnotationTypes_1.SoTranscriptClass.instance[c.type] !== undefined;
                 return isTranscript ? (p + 1) : p;
             }, 0);
-            let gene = Object.assign({}, featureCommon, { type: AnnotationTypes_1.GenomeFeatureType.Gene, class: AnnotationTypes_1.SoGeneClass.instance[feature.type], strand: feature.strand, transcriptCount: transcriptCount });
+            let gene = Object.assign(Object.assign({}, featureCommon), { type: AnnotationTypes_1.GenomeFeatureType.Gene, class: AnnotationTypes_1.SoGeneClass.instance[feature.type], strand: feature.strand, transcriptCount: transcriptCount });
             tile.content.push(gene);
         }
         else if (AnnotationTypes_1.SoTranscriptClass.instance[feature.type] !== undefined) {
             // is transcript	
-            let transcript = Object.assign({}, featureCommon, { type: AnnotationTypes_1.GenomeFeatureType.Transcript, class: AnnotationTypes_1.SoTranscriptClass.instance[feature.type] });
+            let transcript = Object.assign(Object.assign({}, featureCommon), { type: AnnotationTypes_1.GenomeFeatureType.Transcript, class: AnnotationTypes_1.SoTranscriptClass.instance[feature.type] });
             tile.content.push(transcript);
         }
         else if (AnnotationTypes_1.SoTranscriptComponentClass.instance[feature.type] !== undefined) {
             // is transcript component	
-            let info = Object.assign({}, featureCommon, { type: AnnotationTypes_1.GenomeFeatureType.TranscriptComponent, class: AnnotationTypes_1.SoTranscriptComponentClass.instance[feature.type] });
+            let info = Object.assign(Object.assign({}, featureCommon), { type: AnnotationTypes_1.GenomeFeatureType.TranscriptComponent, class: AnnotationTypes_1.SoTranscriptComponentClass.instance[feature.type] });
             if (feature.phase != null) {
                 info.phase = feature.phase;
             }

tools/genome-preprocess/dist/src/track/annotation/AnnotationTypes.js → tools/genome-preprocess/dist/tools/genome-preprocess/src/gff3/AnnotationTypes.js

@@ -1,5 +1,6 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
+exports.SoTranscriptComponentClass = exports.SoTranscriptClass = exports.SoGeneClass = exports.TranscriptComponentClass = exports.TranscriptClass = exports.GeneClass = exports.GenomeFeatureType = exports.Strand = void 0;
 var Strand;
 (function (Strand) {
     Strand["None"] = ".";
@@ -50,8 +51,8 @@ class SoGeneClass {
         this['pseudogene'] = GeneClass.Pseudo;
     }
 }
-SoGeneClass.instance = new SoGeneClass();
 exports.SoGeneClass = SoGeneClass;
+SoGeneClass.instance = new SoGeneClass();
 class SoTranscriptClass {
     constructor() {
         this['transcript'] = TranscriptClass.Unspecified;
@@ -66,8 +67,8 @@ class SoTranscriptClass {
         this['snRNA'] = TranscriptClass.NonProteinCoding;
     }
 }
-SoTranscriptClass.instance = new SoTranscriptClass();
 exports.SoTranscriptClass = SoTranscriptClass;
+SoTranscriptClass.instance = new SoTranscriptClass();
 class SoTranscriptComponentClass {
     constructor() {
         this['CDS'] = TranscriptComponentClass.ProteinCodingSequence;
@@ -76,5 +77,5 @@ class SoTranscriptComponentClass {
         this['three_prime_UTR'] = TranscriptComponentClass.Untranslated;
     }
 }
-SoTranscriptComponentClass.instance = new SoTranscriptComponentClass();
 exports.SoTranscriptComponentClass = SoTranscriptComponentClass;
+SoTranscriptComponentClass.instance = new SoTranscriptComponentClass();

tools/genome-preprocess/dist/tools/genome-preprocess/src/gff3/Convert.js

@@ -16,6 +16,7 @@
  *  - A 'gene' by default is a protein-coding gene
  */
 Object.defineProperty(exports, "__esModule", { value: true });
+exports.gff3Convert = void 0;
 const fs = require("fs");
 const path = require("path");
 const AnnotationTileset_1 = require("./AnnotationTileset");

tools/genome-preprocess/dist/tools/genome-preprocess/src/gff3/Util.js

@@ -1,5 +1,6 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
+exports.getBranches = exports.printSummary = void 0;
 const Terminal_1 = require("../Terminal");
 function printSummary(features) {
     Terminal_1.default.log('Parsing complete\n');

tools/genome-preprocess/dist/tools/genome-preprocess/src/vcf/Convert.js

@@ -1,5 +1,6 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
+exports.vcfConvert = void 0;
 const fs = require("fs");
 const path = require("path");
 const Terminal_1 = require("../Terminal");
@@ -37,11 +38,10 @@ function vcfConvert(inputFilePath, outputDirectory) {
                         alts: feature.ALT.split('')
                     });
                 }
-                // @! temporary for demo
-                let species = 'l_fortunei';
-                saveSequence(tileset.sequences['main'] || [], `${outputDirectory}/${species.toLowerCase()}.vvariants-dir/${species.toLowerCase()}`);
+                let inputFilename = path.basename(inputFilePath);
+                saveSequence(tileset.sequences['main'] || [], `${outputDirectory}/${inputFilename.toLowerCase()}.vvariants-dir/${inputFilename.toLowerCase()}`);
                 // @! temporary, save out genes for biobureau demo
-                filesWritten = filesWritten.concat(biobureauGenerateGenes(inputFilePath, outputDirectory, vcf));
+                // filesWritten = filesWritten.concat(biobureauGenerateGenes(inputFilePath, outputDirectory, vcf));
                 resolve(filesWritten);
             }
         });
@@ -255,7 +255,7 @@ class VCFParser {
             }
             this.callbacks.onComplete(this.output);
         };
-        this.callbacks = Object.assign({}, this.callbacks, callbacks);
+        this.callbacks = Object.assign(Object.assign({}, this.callbacks), callbacks);
     }
     onMetaLine(line) {
         if (line.trim() === '')

tools/genome-preprocess/src/gff3/AnnotationTileset.ts

@@ -1,5 +1,5 @@
 import { Feature } from "genomics-formats/lib/gff3/Feature";
-import { SoGeneClass, SoTranscriptClass, GenomeFeature, GenomeFeatureType, SoTranscriptComponentClass, GeneInfo, TranscriptComponentInfo, TranscriptInfo } from "../../../../src/track/annotation/AnnotationTypes";
+import { SoGeneClass, SoTranscriptClass, GenomeFeature, GenomeFeatureType, SoTranscriptComponentClass, GeneInfo, TranscriptComponentInfo, TranscriptInfo } from "./AnnotationTypes";
 
 export type AnnotationTile = {
     startIndex: number,

tools/genome-preprocess/src/gff3/AnnotationTypes.ts

@@ -0,0 +1,118 @@
+export enum Strand {
+	None = '.',
+	Unknown = '?',
+	Positive = '+',
+	Negative = '-'
+}
+
+export enum GenomeFeatureType {
+	// order corresponds to nesting depth
+	Gene,
+	Transcript,
+	TranscriptComponent,
+}
+
+export interface GenomeFeature {
+	type: GenomeFeatureType,
+}
+
+export enum GeneClass {
+	// this is a small, simplified subset of types specified in the Sequence Ontology
+	Unspecified,
+	ProteinCoding, // assumed default
+	NonProteinCoding, // aka regulatory
+	Pseudo, // non-functional imperfect copy
+}
+
+export interface GeneInfo extends GenomeFeature {
+	type: GenomeFeatureType.Gene,
+	name?: string,
+	startIndex: number,
+	length: number,
+	strand: Strand,
+	class: GeneClass,
+	soClass: keyof SoGeneClass,
+	transcriptCount: number,
+	score?: number,
+}
+
+export enum TranscriptClass {
+	Unspecified,
+	// aka protein coding RNA
+	ProteinCoding,
+	// non-protein coding
+	NonProteinCoding,
+		// sub-types include
+		// Ribosomal
+		// Transfer
+		// Small nuclear
+		// Small nucleolar
+}
+
+/**
+ * Mature transcript – transcript after processing
+ */
+export interface TranscriptInfo extends GenomeFeature {
+	type: GenomeFeatureType.Transcript,
+	name?: string,
+	startIndex: number,
+	length: number,
+	class: TranscriptClass,
+	soClass: keyof SoTranscriptClass,
+}
+
+export enum TranscriptComponentClass {
+	Exon,
+	Untranslated,
+	ProteinCodingSequence,
+}
+
+export interface TranscriptComponentInfo extends GenomeFeature {
+	type: GenomeFeatureType.TranscriptComponent,
+	name?: string,
+	startIndex: number,
+	length: number,
+	class: TranscriptComponentClass,
+	soClass: keyof SoTranscriptComponentClass,
+	phase?: number, // see https://github.com/The-Sequence-Ontology/Specifications/blob/master/gff3.md#description-of-the-format
+}
+
+// small sub set of SO terms found in the Ensemble gff3 files
+// for a more complete set, we should use data from https://github.com/The-Sequence-Ontology/SO-Ontologies
+export class SoGeneClass {
+	[key: string]: undefined | GeneClass;
+
+	readonly 'gene' = GeneClass.Unspecified;
+	readonly 'ncRNA_gene' = GeneClass.NonProteinCoding;
+	readonly 'pseudogene' = GeneClass.Pseudo;
+
+	static readonly instance = new SoGeneClass();
+}
+
+export class SoTranscriptClass {
+	[key: string]: undefined | TranscriptClass;
+
+	readonly 'transcript' = TranscriptClass.Unspecified;
+	readonly 'lnc_RNA' = TranscriptClass.NonProteinCoding;
+	readonly 'mRNA' = TranscriptClass.ProteinCoding;
+	readonly 'pseudogenic_transcript' = TranscriptClass.Unspecified;
+	readonly 'miRNA' = TranscriptClass.NonProteinCoding;
+	readonly 'ncRNA' = TranscriptClass.NonProteinCoding;
+	readonly 'rRNA' = TranscriptClass.NonProteinCoding;
+	readonly 'scRNA' = TranscriptClass.NonProteinCoding;
+	readonly 'snoRNA' = TranscriptClass.NonProteinCoding;
+	readonly 'snRNA' = TranscriptClass.NonProteinCoding;
+
+	static readonly instance = new SoTranscriptClass();
+}
+
+export class SoTranscriptComponentClass {
+	[key: string]: undefined | TranscriptComponentClass;
+
+	readonly 'CDS' = TranscriptComponentClass.ProteinCodingSequence;
+	readonly 'exon' = TranscriptComponentClass.Exon;
+	readonly 'five_prime_UTR' = TranscriptComponentClass.Untranslated;
+	readonly 'three_prime_UTR' = TranscriptComponentClass.Untranslated;
+
+	static readonly instance = new SoTranscriptComponentClass();
+}

tools/genome-preprocess/src/vcf/Convert.ts

@@ -43,12 +43,11 @@ export function vcfConvert(inputFilePath: string, outputDirectory: string): Prom
                     });
                 }
 
-                // @! temporary for demo
-                let species = 'l_fortunei';
-                saveSequence(tileset.sequences['main'] || [], `${outputDirectory}/${species.toLowerCase()}.vvariants-dir/${species.toLowerCase()}`);
+                let inputFilename = path.basename(inputFilePath);
+                saveSequence(tileset.sequences['main'] || [], `${outputDirectory}/${inputFilename.toLowerCase()}.vvariants-dir/${inputFilename.toLowerCase()}`);
 
                 // @! temporary, save out genes for biobureau demo
-                filesWritten = filesWritten.concat(biobureauGenerateGenes(inputFilePath, outputDirectory, vcf));
+                // filesWritten = filesWritten.concat(biobureauGenerateGenes(inputFilePath, outputDirectory, vcf));
 
                 resolve(filesWritten);
             }
@@ -130,7 +129,7 @@ function biobureauGenerateGenes(inputFilePath: string, outputDirectory: string,
         throw `Biobureau demo: filename does not match (@! remove this)`;
     }
 
-    let filesWritten = new Set();
+    let filesWritten = new Set<string>();
 
     let biobureauGeneTileset = new AnnotationTileset(
         lodLevel0TileSize, // ~1 million,