@adntro/raw-dna-dtc-parser - npm Package Compare versions

Comparing version 0.0.12 to 0.1.0

build/src/raw.converter.d.ts

 import { ValidationInfo } from './raw.models';
 import { RawFormatNormalizerTransform } from './raw.normalizer';
-export declare function convertLocalDtcFile(filepath: string, outfilepath?: string, gzipOutput?: boolean, rawTransformer?: RawFormatNormalizerTransform): Promise<{
+export declare function convertLocalDtcFile(filepath: string, outfilepath?: string, gzipOutput?: boolean, rawTransformer?: RawFormatNormalizerTransform, highWaterMark?: number): Promise<{
     validation: ValidationInfo;
     normalizedLocalFile: string;
 }>;

build/src/raw.converter.js

@@ -5,10 +5,10 @@ "use strict";
 const fs_1 = require("fs");
-const zlib_1 = require("zlib");
 const split2 = require("split2");
 const raw_normalizer_1 = require("./raw.normalizer");
-const stream_1 = require("stream");
-function convertLocalDtcFile(filepath, outfilepath = 'converted.txt', gzipOutput = false, rawTransformer = new raw_normalizer_1.RawFormatNormalizerTransform()) {
-    const rs = fs_1.createReadStream(filepath, 'utf-8');
-    const normalizedLocalFile = `${outfilepath}${gzipOutput && outfilepath.indexOf('.gz') === -1 ? '.gz' : ''}`;
-    const out = fs_1.createWriteStream(normalizedLocalFile);
+const files_1 = require("./utils/files");
+function convertLocalDtcFile(filepath, outfilepath = 'converted.txt', gzipOutput = false, rawTransformer = new raw_normalizer_1.RawFormatNormalizerTransform(), highWaterMark = 256 * 1024) {
+    const rs = fs_1.createReadStream(filepath, { encoding: 'utf-8', highWaterMark });
+    const normalizedLocalFile = `${outfilepath}${false && gzipOutput && outfilepath.indexOf('.gz') === -1 ? '.gz' : ''}`;
+    //const normalizedLocalFile = `${outfilepath}`;
+    const out = fs_1.createWriteStream(normalizedLocalFile, { encoding: 'utf-8', highWaterMark });
     const p = new Promise((resolve, reject) => {
@@ -18,3 +18,3 @@ rs
             .pipe(rawTransformer)
-            .pipe(gzipOutput ? zlib_1.createGzip() : new stream_1.PassThrough())
+            //   .pipe(gzipOutput ? createGzip() : new PassThrough())
             .pipe(out);
@@ -28,5 +28,12 @@ rs.on('error', err => reject(err));
     p.finally(() => rs.close());
-    return p;
+    return p.then(({ validation, normalizedLocalFile }) => {
+        if (gzipOutput) {
+            return files_1.gzip(normalizedLocalFile).then(normalizedLocalFileGz => ({ validation, normalizedLocalFile: normalizedLocalFileGz }));
+        }
+        else {
+            return { validation, normalizedLocalFile };
+        }
+    });
 }
 exports.convertLocalDtcFile = convertLocalDtcFile;
 //# sourceMappingURL=raw.converter.js.map

build/src/raw.normalizer.d.ts

@@ -19,2 +19,3 @@ /// <reference types="node" />
     debug: boolean;
+    checkBuild: boolean;
 }
@@ -26,2 +27,3 @@ export declare class RawFormatNormalizerTransform extends Transform {
     debug: boolean;
+    checkBuild: boolean;
     warnings: Set<string>;
@@ -28,0 +30,0 @@ lineCount: number;

build/src/raw.normalizer.js

@@ -50,2 +50,3 @@ "use strict";
         this.debug = false;
+        this.checkBuild = false;
         this.warnings = new Set();
@@ -69,2 +70,4 @@ this.lineCount = 0;
             this.debug = true;
+        if ((opts === null || opts === void 0 ? void 0 : opts.checkBuild) === true)
+            this.checkBuild = true;
         this.on(exports.EVENTS.HEADER, header => {
@@ -85,6 +88,6 @@ this.format = guessFormat(header);
             //emit header
-            this.push('#rsid\tchr\tposition\tbase\n');
+            this.push('#rsid\tchromosome\tposition\tgenotype\n');
         }
         this.lineCount++;
-        const str = String(chunk);
+        const str = typeof chunk === 'string' ? chunk : String(chunk);
         if (!this.genotypeStarted) {
@@ -102,2 +105,3 @@ if (str[0] === '#' || str.match(/rsid|chromosome|chr/i) !== null) {
                 const snp = raw_line_parser_1.convertLine2Snp(str);
+                //const snp: Snp = { chr: '1', position: 1, rsid: 'rskjo', nocall: false, a1: 'A', a2: 'T'};
                 if (snp.chr === 'XY')
@@ -133,7 +137,9 @@ throw new Error('Skip pseudoautosomal XY');
                 // build 37 38
-                const build = genome_build_1.checkBuildForSnp(snp);
-                if (build === 'b37')
-                    this.snpInfo.b37++;
-                else if (build === 'b38')
-                    this.snpInfo.b38++;
+                if (this.checkBuild) {
+                    const build = genome_build_1.checkBuildForSnp(snp);
+                    if (build === 'b37')
+                        this.snpInfo.b37++;
+                    else if (build === 'b38')
+                        this.snpInfo.b38++;
+                }
             }
@@ -161,11 +167,13 @@ catch (e) {
         }
-        if (this.snpInfo.b37 > 0 && this.snpInfo.b38 > 0) {
-            errors.push(raw_errors_1.ERROR_GENOME_BUILD_MIX);
+        if (this.checkBuild) {
+            if (this.snpInfo.b37 > 0 && this.snpInfo.b38 > 0) {
+                errors.push(raw_errors_1.ERROR_GENOME_BUILD_MIX);
+            }
+            else if (this.snpInfo.b37 === 0 && this.snpInfo.b38 === 0) {
+                errors.push(raw_errors_1.ERROR_GENOME_BUILD_NOT_DETECTED);
+            }
+            else if (this.snpInfo.b37 < 25 && this.snpInfo.b38 < 25) {
+                errors.push(raw_errors_1.ERROR_GENOME_BUILD_NOT_ENOUGH);
+            }
         }
-        else if (this.snpInfo.b37 === 0 && this.snpInfo.b38 === 0) {
-            errors.push(raw_errors_1.ERROR_GENOME_BUILD_NOT_DETECTED);
-        }
-        else if (this.snpInfo.b37 < 25 && this.snpInfo.b38 < 25) {
-            errors.push(raw_errors_1.ERROR_GENOME_BUILD_NOT_ENOUGH);
-        }
         if (this.snpInfo.hetRatio > 60 || this.snpInfo.hetRatio < 10) {
@@ -172,0 +180,0 @@ errors.push(raw_errors_1.ERROR_HET_RATIO);

build/src/utils/raw-line-parser.d.ts

 import { Snp } from '../raw.models';
 export declare function cleanGenotypeLine(line: string): string;
 /**
- *   mapper['X'] = 23
+ * 2010-08-Y-3314,0,0,--
+ * rs4475691       1       846808  CC
+ * "rs11240777","1","798959","AG"
+ * rs199474699	26	15990	C	C
+ */
+export declare function convertLine2Snp(line: string): Snp;
+/**
+  mapper['X'] = 23
   mapper['Y'] = 24
@@ -9,2 +16,2 @@ mapper['XY'] = 25
  */
-export declare function convertLine2Snp(line: string): Snp;
+export declare function convertLine2SnpGeneric(line: string): Snp;

build/src/utils/raw-line-parser.js

 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.convertLine2Snp = exports.cleanGenotypeLine = void 0;
+exports.convertLine2SnpGeneric = exports.convertLine2Snp = exports.cleanGenotypeLine = void 0;
 const snp_utils_1 = require("./snp-utils");
@@ -10,3 +10,13 @@ function cleanGenotypeLine(line) {
 /**
- *   mapper['X'] = 23
+ * 2010-08-Y-3314,0,0,--
+ * rs4475691       1       846808  CC
+ * "rs11240777","1","798959","AG"
+ * rs199474699	26	15990	C	C
+ */
+function convertLine2Snp(line) {
+    return convertLine2SnpGeneric(line);
+}
+exports.convertLine2Snp = convertLine2Snp;
+/**
+  mapper['X'] = 23
   mapper['Y'] = 24
@@ -16,3 +26,3 @@ mapper['XY'] = 25
  */
-function convertLine2Snp(line) {
+function convertLine2SnpGeneric(line) {
     if (('' + line).length > 60)
@@ -80,3 +90,3 @@ throw new Error('invalid snp line length');
 }
-exports.convertLine2Snp = convertLine2Snp;
+exports.convertLine2SnpGeneric = convertLine2SnpGeneric;
 //# sourceMappingURL=raw-line-parser.js.map

package.json

 {
   "name": "@adntro/raw-dna-dtc-parser",
-  "version": "0.0.12",
+  "version": "0.1.0",
   "description": "DNA file reader (in DTC formats) and parser",
@@ -5,0 +5,0 @@ "author": "Adntro Genetics SL",

No alert changes

Improved metrics

Total package byte prevSize

60607

increased by12.87%

Number of package files

33

increased by10%

Lines of code

1017

increased by12.13%

Worsened metrics

Number of low supply chain risk alerts

2

increased by100%

No dependency changes