Revert parseVariantDecorator and use jsonifyVariant instead

bcgsc · Jan 3, 2025 · b70cc9a · b70cc9a
1 parent e23f2d0
commit b70cc9a
Show file tree

Hide file tree

Showing 12 changed files with 45 additions and 117 deletions.
diff --git a/src/PMC4232638/index.js b/src/PMC4232638/index.js
@@ -1,16 +1,12 @@
 const readXlsxFile = require('read-excel-file/node');
 
-const { parseVariant: parseVariantOriginal } = require('@bcgsc-pori/graphkb-parser');
+const { jsonifyVariant, parseVariant } = require('@bcgsc-pori/graphkb-parser');
 const { logger } = require('../logging');
 const { rid } = require('../graphkb');
 const _pubmed = require('../entrez/pubmed');
 const _entrezGene = require('../entrez/gene');
 const { PMC4232638: SOURCE_DEFN } = require('../sources');
 
-const { parseVariantDecorator } = require('../util');
-
-const parseVariant = parseVariantDecorator(parseVariantOriginal);
-
 const TP53_COLS = {
     DOM: 'Functional categories for TP53 - Dominant negative activity',
     GOF: 'Functional categories for TP53 - Gain of function',
@@ -137,7 +133,7 @@ const uploadFile = async ({ conn, filename }) => {
         logger.info(`loading: ${row.Gene}:${row['Amino acid change']}`);
 
         try {
-            const parsed = parseVariant(`p.${row['Amino acid change']}`, false).toJSON();
+            const parsed = jsonifyVariant(parseVariant(`p.${row['Amino acid change']}`, false));
             const [gene] = await _entrezGene.fetchAndLoadBySymbol(conn, row.Gene);
             const relevance = await conn.getVocabularyTerm(row.relevance);
             const evidence = await _pubmed.fetchAndLoadByIds(conn, row.evidence);

diff --git a/src/cancergenomeinterpreter/index.js b/src/cancergenomeinterpreter/index.js
@@ -1,9 +1,6 @@
 const fs = require('fs');
 
-const { parseVariant: parseVariantOriginal } = require('@bcgsc-pori/graphkb-parser');
-const { parseVariantDecorator } = require('../util');
-
-const parseVariant = parseVariantDecorator(parseVariantOriginal);
+const { jsonifyVariant, parseVariant } = require('@bcgsc-pori/graphkb-parser');
 
 const {
     loadDelimToJson,
@@ -220,7 +217,7 @@ const processVariants = async ({ conn, row, source }) => {
     }
 
     if (genomic && !isCat) {
-        const parsed = parseVariant(genomic).toJSON();
+        const parsed = jsonifyVariant(parseVariant(genomic));
         const reference1 = await conn.getUniqueRecordBy({
             filters: {
                 AND: [
@@ -245,7 +242,7 @@ const processVariants = async ({ conn, row, source }) => {
     }
 
     if (protein && !isCat) {
-        const parsed = parseVariant(`${gene}:${protein.split(':')[1]}`).toJSON();
+        const parsed = jsonifyVariant(parseVariant(`${gene}:${protein.split(':')[1]}`));
         const type = await conn.getVocabularyTerm(parsed.type);
         proteinVariant = await conn.addVariant({
             content: { ...parsed, reference1: rid(gene1Record), type },
@@ -254,7 +251,7 @@ const processVariants = async ({ conn, row, source }) => {
         });
     }
     if (transcript && cds && !isCat) {
-        const parsed = parseVariant(`${transcript}:${cds}`).toJSON();
+        const parsed = jsonifyVariant(parseVariant(`${transcript}:${cds}`));
         const reference1 = await conn.getUniqueRecordBy({
             filters: { AND: [{ biotype: 'transcript' }, { sourceId: transcript }, { sourceIdVersion: null }] },
             sort: orderPreferredOntologyTerms,
@@ -268,7 +265,7 @@ const processVariants = async ({ conn, row, source }) => {
         });
     }
     if (exonic && !isCat) {
-        const parsed = parseVariant(`${gene}:${exonic}`).toJSON();
+        const parsed = jsonifyVariant(parseVariant(`${gene}:${exonic}`));
         const type = await conn.getVocabularyTerm(parsed.type);
         exonicVariant = await conn.addVariant({
             content: { ...parsed, reference1: rid(gene1Record), type },

diff --git a/src/cancerhotspots/index.js b/src/cancerhotspots/index.js
@@ -5,10 +5,7 @@ const fs = require('fs');
 
 const csv = require('fast-csv');
 
-const { parseVariant: parseVariantOriginal } = require('@bcgsc-pori/graphkb-parser');
-const { parseVariantDecorator } = require('../util');
-
-const parseVariant = parseVariantDecorator(parseVariantOriginal);
+const { jsonifyVariant, parseVariant } = require('@bcgsc-pori/graphkb-parser');
 
 const {
     convertRowFields,
@@ -104,7 +101,7 @@ const processVariants = async ({ conn, record, source }) => {
             // deletion
             notation = `${notation}${start}_${stop}del${refSeq}`;
         }
-        const variant = parseVariant(notation).toJSON();
+        const variant = jsonifyVariant(parseVariant(notation));
 
         variant.reference1 = rid(reference1);
         variant.type = rid(await conn.getVocabularyTerm(variant.type));
@@ -128,10 +125,10 @@ const processVariants = async ({ conn, record, source }) => {
             [reference1] = await _entrezGene.fetchAndLoadByIds(conn, [geneId]);
             featureCache[geneId] = reference1;
         }
-        const variant = parseVariant(
+        const variant = jsonifyVariant(parseVariant(
             protein.replace(/fs\*\?$/, 'fs'), // ignore uncertain truncations
             false,
-        ).toJSON();
+        ));
         variant.reference1 = rid(reference1);
         variant.type = rid(await conn.getVocabularyTerm(variant.type));
         proteinVariant = rid(await conn.addVariant({
@@ -166,7 +163,7 @@ const processVariants = async ({ conn, record, source }) => {
             featureCache[transcriptId] = reference1;
         }
         // parse the cds variant
-        const variant = parseVariant(cds, false).toJSON();
+        const variant = jsonifyVariant(parseVariant(cds, false));
 
         variant.reference1 = reference1;
         variant.type = rid(await conn.getVocabularyTerm(variant.type));

diff --git a/src/cgl/index.js b/src/cgl/index.js
@@ -1,6 +1,6 @@
 const fs = require('fs');
 
-const { parseVariant } = require('@bcgsc-pori/graphkb-parser');
+const { jsonifyVariant, parseVariant } = require('@bcgsc-pori/graphkb-parser');
 
 const {
     loadDelimToJson,
@@ -50,7 +50,7 @@ const loadCdsVariant = async (graphkbConn, transcriptId, cdsNotation) => {
     variant.reference1 = reference1;
     variant.type = rid(await graphkbConn.getVocabularyTerm(variant.type));
     const cds = rid(await graphkbConn.addVariant({
-        content: { ...variant },
+        content: { ...jsonifyVariant(variant) },
         existsOk: true,
         target: 'PositionalVariant',
     }));
@@ -91,7 +91,7 @@ const loadProteinVariant = async (graphkbConn, gene, proteinNotation) => {
     variant.reference1 = reference1;
     variant.type = rid(await graphkbConn.getVocabularyTerm(variant.type));
     const protein = rid(await graphkbConn.addVariant({
-        content: { ...variant },
+        content: { ...jsonifyVariant(variant) },
         existsOk: true,
         target: 'PositionalVariant',
     }));
@@ -167,7 +167,7 @@ const loadGenomicVariant = async (graphkbConn, chromosome, position, ref, alt) =
     variant.reference1 = reference1;
     variant.type = rid(await graphkbConn.getVocabularyTerm(variant.type));
     const genomic = rid(await graphkbConn.addVariant({
-        content: { ...variant, assembly: 'hg19' },
+        content: { ...jsonifyVariant(variant), assembly: 'hg19' },
         existsOk: true,
         target: 'PositionalVariant',
     }));

diff --git a/src/civic/variant.js b/src/civic/variant.js
@@ -1,14 +1,16 @@
-const { ErrorMixin, ParsingError, parseVariant: parseVariantOriginal } = require('@bcgsc-pori/graphkb-parser');
+const {
+    ErrorMixin,
+    jsonifyVariant,
+    parseVariant,
+    ParsingError,
+} = require('@bcgsc-pori/graphkb-parser');
+
 const { rid } = require('../graphkb');
 const _entrezGene = require('../entrez/gene');
 const _snp = require('../entrez/snp');
 const { civic: SOURCE_DEFN } = require('../sources');
 const { logger } = require('../logging');
 
-const { parseVariantDecorator } = require('../util');
-
-const parseVariant = parseVariantDecorator(parseVariantOriginal);
-
 class NotImplementedError extends ErrorMixin { }
 
 const VARIANT_CACHE = new Map();
@@ -523,7 +525,7 @@ const uploadVariant = async (conn, normalizedVariant) => {
     let content = {};
 
     if (normalizedVariant.positional) {
-        content = parseVariant(normalizedVariant.variant, false).toJSON();
+        content = jsonifyVariant(parseVariant(normalizedVariant.variant, false));
     }
 
     // Variant type

diff --git a/src/cosmic/resistance.js b/src/cosmic/resistance.js
@@ -3,10 +3,7 @@
  */
 const fs = require('fs');
 
-const { parseVariant: parseVariantOriginal } = require('@bcgsc-pori/graphkb-parser');
-const { parseVariantDecorator } = require('../util');
-
-const parseVariant = parseVariantDecorator(parseVariantOriginal);
+const { jsonifyVariant, parseVariant } = require('@bcgsc-pori/graphkb-parser');
 
 const {
     loadDelimToJson,
@@ -80,7 +77,7 @@ const processVariants = async ({ conn, record, source }) => {
 
     try {
         // add the protein variant with its protein translation
-        const variant = parseVariant(record.protein, false).toJSON();
+        const variant = jsonifyVariant(parseVariant(record.protein, false));
         variant.type = rid(await conn.getVocabularyTerm(variant.type));
 
         const reference1 = rid(await _ensembl.fetchAndLoadById(
@@ -116,7 +113,7 @@ const processVariants = async ({ conn, record, source }) => {
     // create the cds variant
     if (record.cds && record.cds.trim()) {
         try {
-            const variant = parseVariant(record.cds, false).toJSON();
+            const variant = jsonifyVariant(parseVariant(record.cds, false));
             // get the ensembl transcript
             const reference1 = rid(await _ensembl.fetchAndLoadById(
                 conn,
@@ -146,7 +143,7 @@ const processVariants = async ({ conn, record, source }) => {
     // add the genomic representation
     if (record.genomic) {
         try {
-            const variant = parseVariant(record.genomic, false).toJSON();
+            const variant = jsonifyVariant(parseVariant(record.genomic, false));
             // get the chromosome
             const reference1 = rid(await conn.getUniqueRecordBy({
                 filters: {

diff --git a/src/docm/index.js b/src/docm/index.js
@@ -6,10 +6,7 @@
 const Ajv = require('ajv');
 const fs = require('fs');
 
-const { parseVariant: parseVariantOriginal } = require('@bcgsc-pori/graphkb-parser');
-const { parseVariantDecorator } = require('../util');
-
-const parseVariant = parseVariantDecorator(parseVariantOriginal);
+const { jsonifyVariant, parseVariant } = require('@bcgsc-pori/graphkb-parser');
 
 const { checkSpec, request } = require('../util');
 const {
@@ -110,7 +107,7 @@ const processVariants = async ({ conn, source, record: docmRecord }) => {
     try {
         // create the protein variant
         const [reference1] = await _gene.fetchAndLoadBySymbol(conn, gene);
-        let variant = parseVariant(parseDocmVariant(aminoAcid), false).toJSON();
+        let variant = jsonifyVariant(parseVariant(parseDocmVariant(aminoAcid), false));
         const type = await conn.getVocabularyTerm(variant.type);
         protein = variant = await conn.addVariant({
             content: { ...variant, reference1: rid(reference1), type: rid(type) },
@@ -124,7 +121,7 @@ const processVariants = async ({ conn, source, record: docmRecord }) => {
 
     try {
         // create the genomic variant
-        let variant = parseVariant(buildGenomicVariant(docmRecord), false).toJSON();
+        let variant = jsonifyVariant(parseVariant(buildGenomicVariant(docmRecord), false));
         const type = await conn.getVocabularyTerm(variant.type);
         const reference1 = await conn.getUniqueRecordBy({
             filters: {

diff --git a/src/entrez/snp.js b/src/entrez/snp.js
@@ -1,6 +1,6 @@
 const Ajv = require('ajv');
 
-const { parseVariant: parseVariantOriginal } = require('@bcgsc-pori/graphkb-parser');
+const { jsonifyVariant, parseVariant } = require('@bcgsc-pori/graphkb-parser');
 const { checkSpec } = require('../util');
 const {
     fetchByIdList, uploadRecord, preLoadCache: preLoadAnyCache, BASE_FETCH_URL,
@@ -10,10 +10,6 @@ const entrezGene = require('./gene');
 const { rid } = require('../graphkb');
 const { logger } = require('../logging');
 
-const { parseVariantDecorator } = require('../util');
-
-const parseVariant = parseVariantDecorator(parseVariantOriginal);
-
 const ajv = new Ajv();
 
 const { dbSnp: SOURCE_DEFN } = require('../sources');
@@ -52,7 +48,7 @@ const loadFromDocsumHgvs = async (api, hgvsVariants) => {
 
     try {
         if (hgvsVariants.cds) {
-            const parsed = parseVariant(hgvsVariants.cds.split('|')[0], true).toJSON();
+            const parsed = jsonifyVariant(parseVariant(hgvsVariants.cds.split('|')[0], true));
             const [transcript] = await refseq.fetchAndLoadByIds(api, [parsed.reference1]);
             const type = await api.getVocabularyTerm(parsed.type);
             cds = await api.addVariant({
@@ -68,7 +64,7 @@ const loadFromDocsumHgvs = async (api, hgvsVariants) => {
     try {
         if (hgvsVariants.protein) {
             const gene = hgvsVariants.protein.split('|').find(p => p.startsWith('GENE='));
-            const parsed = parseVariant(hgvsVariants.protein.split('|')[0], true).toJSON();
+            const parsed = jsonifyVariant(parseVariant(hgvsVariants.protein.split('|')[0], true));
             const [reference1] = await refseq.fetchAndLoadByIds(api, [parsed.reference1]);
             const type = await api.getVocabularyTerm(parsed.type);
             protein = await api.addVariant({

diff --git a/src/moa/index.js b/src/moa/index.js
@@ -1,9 +1,6 @@
 const Ajv = require('ajv');
 
-const { parseVariant: parseVariantOriginal } = require('@bcgsc-pori/graphkb-parser');
-const { parseVariantDecorator } = require('../util');
-
-const parseVariant = parseVariantDecorator(parseVariantOriginal);
+const { jsonifyVariant, parseVariant } = require('@bcgsc-pori/graphkb-parser');
 
 const { checkSpec, requestWithRetry } = require('../util');
 const { moa: SOURCE_DEFN } = require('../sources');
@@ -62,7 +59,7 @@ const loadSmallMutation = async (conn, gene, moaVariant) => {
 
     // create the genomic variant if we have the appropriate fields
     if (!['reference_allele', 'alternate_allele', 'start_position', 'end_position', 'chromosome'].some(v => moaVariant[v] === null)) {
-        const hgvsg = parseVariant(composeGenomicHgvs(moaVariant), false).toJSON();
+        const hgvsg = jsonifyVariant(parseVariant(composeGenomicHgvs(moaVariant), false));
         const chromosome = await conn.getUniqueRecordBy({
             filters: {
                 AND: [
@@ -88,7 +85,7 @@ const loadSmallMutation = async (conn, gene, moaVariant) => {
 
     // create the cds variant
     if (moaVariant.cdna_change !== null && moaVariant.cdna_change !== '') {
-        const hgvsc = parseVariant(moaVariant.cdna_change, false).toJSON();
+        const hgvsc = jsonifyVariant(parseVariant(moaVariant.cdna_change, false));
         const cdsType = rid(await conn.getVocabularyTerm(hgvsc.type));
         cdsVariant = rid(await conn.addVariant({
             content: {
@@ -101,7 +98,7 @@ const loadSmallMutation = async (conn, gene, moaVariant) => {
 
     // create the protein variant
     if (moaVariant.protein_change !== null && moaVariant.protein_change !== '') {
-        const hgvsp = parseVariant(moaVariant.protein_change, false).toJSON();
+        const hgvsp = jsonifyVariant(parseVariant(moaVariant.protein_change, false));
         const proteinType = rid(await conn.getVocabularyTerm(hgvsp.type));
         proteinVariant = rid(await conn.addVariant({
             content: {
@@ -132,7 +129,7 @@ const loadSmallMutation = async (conn, gene, moaVariant) => {
         } else {
             variantType = await conn.getVocabularyTerm('mutation');
         }
-        const parsed = parseVariant(`e.${exonNumber}mut`, false).toJSON();
+        const parsed = jsonifyVariant(parseVariant(`e.${exonNumber}mut`, false));
         exonVariant = await conn.addVariant({
             content: {
                 ...parsed,

diff --git a/src/oncokb/index.js b/src/oncokb/index.js
@@ -5,10 +5,7 @@ const Ajv = require('ajv');
 const fs = require('fs');
 const path = require('path');
 
-const { ParsingError, parseVariant: parseVariantOriginal } = require('@bcgsc-pori/graphkb-parser');
-const { parseVariantDecorator } = require('../util');
-
-const parseVariant = parseVariantDecorator(parseVariantOriginal);
+const { jsonifyVariant, parseVariant, ParsingError } = require('@bcgsc-pori/graphkb-parser');
 
 const {
     checkSpec,
@@ -279,11 +276,11 @@ const processVariant = async (conn, {
 
     if (!variant) {
         try {
-            variant = parseVariant(type, false).toJSON();
+            variant = jsonifyVariant(parseVariant(type, false));
         } catch (err) {
             try {
                 // try with adding a p prefix also
-                variant = parseVariant(`p.${type}`, false).toJSON();
+                variant = jsonifyVariant(parseVariant(`p.${type}`, false));
             } catch (err2) { }
             logger.warn(`failed to parse the variant (${type}) for record (gene=${gene}, variant=${variantName})`);
             throw err;
@@ -330,7 +327,7 @@ const processVariant = async (conn, {
 
         try {
             // try with adding a p prefix also
-            const parsed = parseVariant(altVariantName, false).toJSON();
+            const parsed = jsonifyVariant(parseVariant(altVariantName, false));
             parsed.reference1 = rid(reference1);
             parsed.type = rid(await getVocabulary(conn, parsed.type));
             const altVariant = rid(await conn.addVariant({