jbrowse-plugin-msaview 2.7.2 → 2.7.3

package/dist/MsaViewPanel/msaCoordToGenomeCoord.js

@@ -1,27 +1,56 @@
+import { getCodonRanges } from 'g2p_mapper';
 import { gappedToUngappedPosition } from './structureConnection';
-export function msaCoordToGenomeCoord({ model, coord: mouseCol, }) {
-    const { querySeqName, transcriptToMsaMap, mafRegion } = model;
-    const querySeq = model.rows.find(f => f[0] === querySeqName)?.[1];
+/**
+ * The genome regions covered by MSA column `coord` of the query row, in 0-based
+ * half-open coordinates (what bpToPx and navTo take).
+ *
+ * Usually one region -- one codon, or one base in a MAF alignment -- but a
+ * codon split across an exon boundary yields one region per contiguous piece,
+ * which is why this returns a list.
+ */
+export function msaCoordToGenomeRegions({ model, coord: mouseCol, }) {
+    const { querySeqName, transcriptToMsaMap, mafRegion, rows } = model;
+    const querySeq = rows.find(f => f[0] === querySeqName)?.[1];
     if (!querySeq) {
-        return undefined;
+        return [];
     }
     const ungappedPos = gappedToUngappedPosition(querySeq, mouseCol);
     if (ungappedPos === undefined) {
-        return undefined;
+        return [];
     }
     if (mafRegion) {
         const genomePos = mafRegion.start + ungappedPos;
         return genomePos < mafRegion.end
-            ? { refName: mafRegion.refName, start: genomePos, end: genomePos + 1 }
-            : undefined;
+            ? [{ refName: mafRegion.refName, start: genomePos, end: genomePos + 1 }]
+            : [];
     }
     if (transcriptToMsaMap) {
-        const { refName, p2g } = transcriptToMsaMap;
-        const s = p2g[ungappedPos];
-        const e = p2g[ungappedPos + 1];
-        return s !== undefined && e !== undefined
-            ? { refName, start: Math.min(s, e), end: Math.max(s, e) }
-            : undefined;
+        const { refName, p2gCodon } = transcriptToMsaMap;
+        // p2gCodon holds every genomic base of the codon, so the range is exact on
+        // either strand. Deriving it from consecutive p2g entries instead
+        // (p2g[pos]..p2g[pos+1]) was off by one base on the reverse strand -- where
+        // p2g stores the codon's *highest* coordinate -- dropped the final residue,
+        // whose successor has no p2g entry, and spanned the whole intron for a
+        // codon split across an exon boundary.
+        return (getCodonRanges(p2gCodon, ungappedPos)?.map(([start, end]) => ({
+            refName,
+            start,
+            end,
+        })) ?? []);
     }
-    return undefined;
+    return [];
+}
+/**
+ * A single region spanning the codon at MSA column `coord`, for navigation. For
+ * a codon split across an exon boundary this spans the intervening intron.
+ */
+export function msaCoordToGenomeCoord(args) {
+    const regions = msaCoordToGenomeRegions(args);
+    const first = regions[0];
+    const last = regions.at(-1);
+    // getCodonRanges returns ranges sorted ascending, so first.start..last.end
+    // bounds the codon
+    return first && last
+        ? { refName: first.refName, start: first.start, end: last.end }
+        : undefined;
 }

package/dist/MsaViewPanel/msaCoordToGenomeCoord.test.js

@@ -1,5 +1,14 @@
+import { genomeToTranscriptSeqMapping } from 'g2p_mapper';
 import { describe, expect, test } from 'vitest';
-import { msaCoordToGenomeCoord } from './msaCoordToGenomeCoord';
+import { msaCoordToGenomeCoord, msaCoordToGenomeRegions, } from './msaCoordToGenomeCoord';
+// codon at protein position i covers three consecutive genome bases starting at
+// 100 + i * 3, i.e. a single-exon forward-strand transcript
+function forwardCodons(n) {
+    return Object.fromEntries(Array.from({ length: n }, (_, i) => [
+        i,
+        [100 + i * 3, 101 + i * 3, 102 + i * 3],
+    ]));
+}
 describe('msaCoordToGenomeCoord', () => {
     test('returns undefined when neither transcriptToMsaMap nor mafRegion is defined', () => {
         const model = {
@@ -16,7 +25,7 @@ describe('msaCoordToGenomeCoord', () => {
             querySeqName: 'QUERY',
             transcriptToMsaMap: {
                 refName: 'chr1',
-                p2g: { 0: 100, 1: 103 },
+                p2gCodon: forwardCodons(2),
             },
             rows: [['OTHER', 'MKAA']],
         };
@@ -28,7 +37,7 @@ describe('msaCoordToGenomeCoord', () => {
             querySeqName: 'QUERY',
             transcriptToMsaMap: {
                 refName: 'chr1',
-                p2g: { 0: 100, 1: 103 },
+                p2gCodon: forwardCodons(2),
             },
             rows: [['QUERY', 'M-KA']],
         };
@@ -41,7 +50,7 @@ describe('msaCoordToGenomeCoord', () => {
             querySeqName: 'QUERY',
             transcriptToMsaMap: {
                 refName: 'chr1',
-                p2g: { 0: 100, 1: 103, 2: 106, 3: 109 },
+                p2gCodon: forwardCodons(4),
             },
             rows: [['QUERY', 'MKAA']],
         };
@@ -58,7 +67,7 @@ describe('msaCoordToGenomeCoord', () => {
             querySeqName: 'QUERY',
             transcriptToMsaMap: {
                 refName: 'chr1',
-                p2g: { 0: 100, 1: 103, 2: 106, 3: 109 },
+                p2gCodon: forwardCodons(4),
             },
             rows: [['QUERY', 'M-K-AA']],
             //                 012345 gapped positions
@@ -79,34 +88,33 @@ describe('msaCoordToGenomeCoord', () => {
             end: 109,
         });
     });
-    test('returns undefined when p2g mapping is incomplete', () => {
+    test('returns undefined when the position has no codon mapping', () => {
         const model = {
             querySeqName: 'QUERY',
             transcriptToMsaMap: {
                 refName: 'chr1',
-                p2g: { 0: 100 }, // Missing entry for position 1
+                p2gCodon: forwardCodons(1),
             },
             rows: [['QUERY', 'MKAA']],
         };
-        // Position 0 needs p2g[0] and p2g[1], but p2g[1] is missing
-        const result = msaCoordToGenomeCoord({ model, coord: 0 });
+        // ungapped position 1 has no entry in p2gCodon
+        const result = msaCoordToGenomeCoord({ model, coord: 1 });
         expect(result).toBeUndefined();
     });
-    test('handles reverse strand (start > end in p2g)', () => {
+    test('maps the final residue, whose codon has no successor', () => {
         const model = {
             querySeqName: 'QUERY',
             transcriptToMsaMap: {
                 refName: 'chr1',
-                p2g: { 0: 109, 1: 106, 2: 103, 3: 100 }, // Reverse strand
+                p2gCodon: forwardCodons(4),
             },
             rows: [['QUERY', 'MKAA']],
         };
-        // Should return min/max correctly
-        const result = msaCoordToGenomeCoord({ model, coord: 0 });
+        const result = msaCoordToGenomeCoord({ model, coord: 3 });
         expect(result).toEqual({
             refName: 'chr1',
-            start: 106, // min(109, 106)
-            end: 109, // max(109, 106)
+            start: 109,
+            end: 112,
         });
     });
     test('returns undefined for out of bounds coord', () => {
@@ -114,7 +122,7 @@ describe('msaCoordToGenomeCoord', () => {
             querySeqName: 'QUERY',
             transcriptToMsaMap: {
                 refName: 'chr1',
-                p2g: { 0: 100, 1: 103 },
+                p2gCodon: forwardCodons(2),
             },
             rows: [['QUERY', 'MK']],
         };
@@ -127,7 +135,7 @@ describe('msaCoordToGenomeCoord', () => {
             querySeqName: 'SEQ2',
             transcriptToMsaMap: {
                 refName: 'chr1',
-                p2g: { 0: 200, 1: 203 },
+                p2gCodon: { 0: [200, 201, 202], 1: [203, 204, 205] },
             },
             rows: [
                 ['SEQ1', 'AAAA'],
@@ -142,6 +150,88 @@ describe('msaCoordToGenomeCoord', () => {
             end: 203,
         });
     });
+    // The mapping comes from the real g2p_mapper rather than hand-written
+    // fixtures: on the reverse strand p2g stores the codon's *highest*
+    // coordinate, which is what the old p2g[pos]..p2g[pos+1] arithmetic got wrong
+    describe('real g2p_mapper mappings', () => {
+        test('forward strand, single exon', () => {
+            const { p2gCodon, refName } = genomeToTranscriptSeqMapping({
+                refName: 'chr1',
+                start: 100,
+                end: 112,
+                strand: 1,
+                subfeatures: [{ refName: 'chr1', type: 'CDS', start: 100, end: 112 }],
+            });
+            const model = {
+                querySeqName: 'QUERY',
+                transcriptToMsaMap: { refName, p2gCodon },
+                rows: [['QUERY', 'MKAA']],
+            };
+            expect(msaCoordToGenomeCoord({ model, coord: 0 })).toEqual({
+                refName: 'chr1',
+                start: 100,
+                end: 103,
+            });
+            expect(msaCoordToGenomeCoord({ model, coord: 3 })).toEqual({
+                refName: 'chr1',
+                start: 109,
+                end: 112,
+            });
+        });
+        test('reverse strand codon covers the last three bases of the CDS', () => {
+            const { p2gCodon, refName } = genomeToTranscriptSeqMapping({
+                refName: 'chr1',
+                start: 100,
+                end: 112,
+                strand: -1,
+                subfeatures: [{ refName: 'chr1', type: 'CDS', start: 100, end: 112 }],
+            });
+            const model = {
+                querySeqName: 'QUERY',
+                transcriptToMsaMap: { refName, p2gCodon },
+                rows: [['QUERY', 'MKAA']],
+            };
+            // the first residue is translated from the 3' end of the genome region
+            expect(msaCoordToGenomeCoord({ model, coord: 0 })).toEqual({
+                refName: 'chr1',
+                start: 109,
+                end: 112,
+            });
+            expect(msaCoordToGenomeCoord({ model, coord: 3 })).toEqual({
+                refName: 'chr1',
+                start: 100,
+                end: 103,
+            });
+        });
+        test('codon split across an exon boundary yields one region per piece', () => {
+            // exon 1 contributes 4 bases, so residue 1 straddles the intron
+            const { p2gCodon, refName } = genomeToTranscriptSeqMapping({
+                refName: 'chr1',
+                start: 100,
+                end: 210,
+                strand: 1,
+                subfeatures: [
+                    { refName: 'chr1', type: 'CDS', start: 100, end: 104 },
+                    { refName: 'chr1', type: 'CDS', start: 200, end: 202 },
+                ],
+            });
+            const model = {
+                querySeqName: 'QUERY',
+                transcriptToMsaMap: { refName, p2gCodon },
+                rows: [['QUERY', 'MK']],
+            };
+            expect(msaCoordToGenomeRegions({ model, coord: 1 })).toEqual([
+                { refName: 'chr1', start: 103, end: 104 },
+                { refName: 'chr1', start: 200, end: 202 },
+            ]);
+            // the single-region form bounds the pieces, for navigation
+            expect(msaCoordToGenomeCoord({ model, coord: 1 })).toEqual({
+                refName: 'chr1',
+                start: 103,
+                end: 202,
+            });
+        });
+    });
     // MAF region tests
     describe('mafRegion', () => {
         test('returns genome position for mafRegion mapping', () => {
@@ -217,7 +307,7 @@ describe('msaCoordToGenomeCoord', () => {
                 querySeqName: 'hg38.chr1',
                 transcriptToMsaMap: {
                     refName: 'chr2',
-                    p2g: { 0: 5000, 1: 5003 },
+                    p2gCodon: { 0: [5000, 5001, 5002], 1: [5003, 5004, 5005] },
                 },
                 mafRegion: {
                     refName: 'chr1',

package/dist/MsaViewPanel/msaDataStore.js

@@ -1,22 +1,13 @@
-import { openDB } from 'idb';
+import { createDbOpener } from '../utils/idb';
 const DB_NAME = 'jbrowse-msaview-data';
 const DB_VERSION = 1;
 const STORE_NAME = 'msa-data';
-let dbPromise;
-function getDB() {
-    dbPromise ??= openDB(DB_NAME, DB_VERSION, {
-        upgrade(db) {
-            if (!db.objectStoreNames.contains(STORE_NAME)) {
-                const store = db.createObjectStore(STORE_NAME, { keyPath: 'id' });
-                store.createIndex('timestamp', 'timestamp', { unique: false });
-            }
-        },
-    }).catch((e) => {
-        dbPromise = undefined;
-        throw e;
-    });
-    return dbPromise;
-}
+const getDB = createDbOpener(DB_NAME, DB_VERSION, db => {
+    if (!db.objectStoreNames.contains(STORE_NAME)) {
+        const store = db.createObjectStore(STORE_NAME, { keyPath: 'id' });
+        store.createIndex('timestamp', 'timestamp', { unique: false });
+    }
+});
 export function generateDataStoreId() {
     return `msa-${Date.now()}-${Math.random().toString(36).slice(2, 11)}`;
 }
@@ -41,7 +32,7 @@ export async function storeMsaData(id, data) {
 export async function retrieveMsaData(id) {
     try {
         const db = await getDB();
-        const result = (await db.get(STORE_NAME, id));
+        const result = await db.get(STORE_NAME, id);
         if (result) {
             return {
                 msa: result.msa,

package/dist/MsaViewPanel/syncGenomeHoverToMsaColumn.test.js

@@ -45,12 +45,14 @@ describe('syncGenomeHoverToMsaColumn (real genomeToMSA mapping)', () => {
     beforeEach(() => {
         vi.clearAllMocks();
     });
-    test('genome hover at coord 1005 highlights MSA column 5', () => {
+    test('genome hover at coord 1005 highlights MSA column 4', () => {
         const { model, calls } = makeModel();
         const run = syncGenomeHoverToMsaColumn(model);
-        hoverGenome(1005); // 1005 - mafRegion.start(1000) = ungapped 5
+        // the hover coord is 1-based, so 1005 is the 0-based base 1004, i.e.
+        // ungapped offset 4 into a region starting at 1000
+        hoverGenome(1005);
         run();
-        expect(calls).toEqual([5]);
+        expect(calls).toEqual([4]);
     });
     test('moving the genome hover moves the highlighted column', () => {
         const { model, calls } = makeModel();
@@ -59,7 +61,7 @@ describe('syncGenomeHoverToMsaColumn (real genomeToMSA mapping)', () => {
         run();
         hoverGenome(1007);
         run();
-        expect(calls).toEqual([2, 7]);
+        expect(calls).toEqual([1, 6]);
     });
     test('leaving the genome clears the column it set', () => {
         const { model, calls } = makeModel();
@@ -68,7 +70,7 @@ describe('syncGenomeHoverToMsaColumn (real genomeToMSA mapping)', () => {
         run();
         clearGenomeHover();
         run();
-        expect(calls).toEqual([4, undefined]);
+        expect(calls).toEqual([3, undefined]);
     });
     test('a hover outside the maf region clears a previously-set column once', () => {
         const { model, calls } = makeModel();
@@ -78,7 +80,7 @@ describe('syncGenomeHoverToMsaColumn (real genomeToMSA mapping)', () => {
         hoverGenome(5000); // outside [1000,1010) -> genomeToMSA returns undefined
         run();
         run();
-        expect(calls).toEqual([4, undefined]);
+        expect(calls).toEqual([3, undefined]);
     });
     test('never touches mouseCol when the genome never provides a column, so a direct MSA hover survives unrelated session hovers', () => {
         const { model, calls } = makeModel();