jbrowse-plugin-msaview 3.3.0 → 3.4.1

package/dist/utils/ncbiOrthologs.js

@@ -53,6 +53,18 @@ function ncbiUrl(url) {
     const sep = url.includes('?') ? '&' : '?';
     return `${url}${sep}tool=${NCBI_TOOL}&email=${encodeURIComponent(NCBI_EMAIL)}`;
 }
+/**
+ * A candidate gene reference as the ortholog services know it: a GFF ID prefix
+ * (gene:TP53) and a version suffix (NM_000546.6) stripped off. Shared with
+ * pantherOrthologs, which asks PANTHER the same question with the same
+ * candidates.
+ */
+export function cleanGeneCandidate(raw) {
+    return raw
+        .trim()
+        .replace(/^\w+:/, '')
+        .replace(/\.\d+$/, '');
+}
 /**
  * A free-text gene reference -> NCBI gene id. A bare number is taken as the id
  * itself; anything else is searched as a gene name within the query taxon.
@@ -69,8 +81,7 @@ export async function resolveGeneId(candidates, taxId) {
         if (/^\d+$/.test(query)) {
             return { geneId: query, matched: query };
         }
-        // strip a version suffix (NM_000546.6) and any GFF ID prefix (gene:TP53)
-        const cleaned = query.replace(/^\w+:/, '').replace(/\.\d+$/, '');
+        const cleaned = cleanGeneCandidate(query);
         const term = `${cleaned}[Gene Name] AND ${taxId}[taxid]`;
         const json = await jsonfetch(ncbiUrl(`${EUTILS}/esearch.fcgi?db=gene&term=${encodeURIComponent(term)}&retmode=json&retmax=1`));
         const geneId = json.esearchresult?.idlist?.[0];
@@ -179,15 +190,23 @@ function sanitize(name) {
  * Sanitized, unique single-token labels used identically in the FASTA headers,
  * the tree leaf names and the domain GFF seq_ids — that identity is how the
  * viewer pairs a tree leaf to its alignment row to its domain track. Collisions
- * get a numeric suffix rather than silently overwriting a row.
+ * get a numeric suffix rather than silently overwriting a row, and the suffix
+ * climbs past labels already emitted: a species genuinely named `human_2` would
+ * otherwise take the label the second `human` is about to get, merging two rows
+ * into one.
  */
 export function dedupeLabels(names) {
-    const seen = new Map();
+    const used = new Set();
     return names.map(name => {
         const base = sanitize(name) || 'row';
-        const n = seen.get(base) ?? 0;
-        seen.set(base, n + 1);
-        return n === 0 ? base : `${base}_${n + 1}`;
+        let label = base;
+        let n = 1;
+        while (used.has(label)) {
+            n += 1;
+            label = `${base}_${n}`;
+        }
+        used.add(label);
+        return label;
     });
 }
 /**

package/dist/utils/ncbiOrthologs.test.js

@@ -1,5 +1,5 @@
 import { afterEach, describe, expect, test, vi } from 'vitest';
-import { dedupeLabels, fetchOrthologGenes, fetchRepresentativeProteins, parseFasta, } from './ncbiOrthologs';
+import { cleanGeneCandidate, dedupeLabels, fetchOrthologGenes, fetchRepresentativeProteins, parseFasta, } from './ncbiOrthologs';
 describe('dedupeLabels', () => {
     test('sanitizes to single tokens', () => {
         // labels are used identically as FASTA headers, Newick leaf names and GFF
@@ -22,6 +22,28 @@ describe('dedupeLabels', () => {
     test('falls back for a name with no usable characters', () => {
         expect(dedupeLabels(['...', '...'])).toEqual(['row', 'row_2']);
     });
+    test('climbs past a suffix a literal name already took', () => {
+        // a label pairs a tree leaf to its alignment row, so handing the same one
+        // to two rows merges two species into one silently. The third label is ugly
+        // and unique, which is the trade
+        expect(dedupeLabels(['human', 'human', 'human_2'])).toEqual([
+            'human',
+            'human_2',
+            'human_2_2',
+        ]);
+        expect(dedupeLabels(['human_2', 'human', 'human'])).toEqual([
+            'human_2',
+            'human',
+            'human_3',
+        ]);
+    });
+});
+describe('cleanGeneCandidate', () => {
+    test('strips what a GFF feature carries that a gene lookup does not want', () => {
+        expect(cleanGeneCandidate(' gene:TP53 ')).toBe('TP53');
+        expect(cleanGeneCandidate('NM_000546.6')).toBe('NM_000546');
+        expect(cleanGeneCandidate('TP53')).toBe('TP53');
+    });
 });
 describe('parseFasta', () => {
     test('keys by the first header token and joins wrapped lines', () => {

package/dist/utils/pantherOrthologs.js

@@ -16,7 +16,7 @@
 // GenPept records with CDD Region features, so the overlay attaches as it does
 // to a RefSeq accession.
 import { jsonfetch } from './fetch';
-import { dedupeLabels, defaultMaxSpecies } from './ncbiOrthologs';
+import { cleanGeneCandidate, dedupeLabels, defaultMaxSpecies, } from './ncbiOrthologs';
 import { fetchTaxonomyInfo } from './taxonomyNames';
 const PANTHER = 'https://pantherdb.org/services/oai/pantherdb';
 const UNIPROT = 'https://rest.uniprot.org/uniprotkb';
@@ -122,14 +122,6 @@ async function fetchSequences(accessions) {
     }
     return map;
 }
-// strip a version suffix (NM_000546.6) and any GFF ID prefix (gene:TP53), as
-// resolveGeneId does for NCBI
-function cleanCandidate(raw) {
-    return raw
-        .trim()
-        .replace(/^\w+:/, '')
-        .replace(/\.\d+$/, '');
-}
 /**
  * One `matchortho` per candidate until PANTHER maps one. A JBrowse feature
  * carries whatever its GFF/BigBed had — `id()`, `name`, `gene_name` — and only
@@ -139,7 +131,7 @@ function cleanCandidate(raw) {
 async function matchOrthologs(candidates, taxId, targets) {
     let matched;
     for (const raw of candidates) {
-        const query = cleanCandidate(raw);
+        const query = cleanGeneCandidate(raw);
         if (!query) {
             continue;
         }

package/dist/utils/phmmer.d.ts

@@ -0,0 +1,54 @@
+import type { PhmmerDatabase } from '../LaunchMsaView/components/BlastQuery/consts';
+export interface PhmmerRow {
+    accession: string;
+    id: string;
+    sciname: string;
+    taxid?: number;
+    title?: string;
+    /** the matched envelope on the target, e.g. '503-912', absent if unparseable */
+    range?: string;
+    /** the row as phmmer aligned it, uppercased with '.' inserts turned into '-' */
+    aligned: string;
+}
+export interface PhmmerAlignment {
+    rows: PhmmerRow[];
+    /**
+     * the query, placed into the same columns. phmmer does not put the query in
+     * its own output, so this is derived — see buildQueryRow.
+     */
+    queryRow: string;
+}
+/**
+ * Human-facing link to a job, shown while it runs and on error.
+ *
+ * The category has to be sss: jdispatcher serves its shell with a 200 for any
+ * category, so /pfa/ and /psa/ look fine to a fetch and render "Page Not Found"
+ * in a browser.
+ */
+export declare function phmmerResultUrl(jobId: string): string;
+/** EBI job ids are prefixed with the tool that made them */
+export declare function isPhmmerJobId(jobId: string): boolean;
+/**
+ * Exported for testing against a captured .sto — the annotation names (RF, the
+ * DE line's OS=/OX=) are the whole risk in this mapping, and nothing else in CI
+ * would notice if HMMER or EBI changed one.
+ */
+export declare function parsePhmmerAlignment({ stockholm, query, }: {
+    stockholm: string;
+    query: string;
+}): PhmmerAlignment;
+export declare function queryPhmmer({ query, database, onProgress, onRid, signal, }: {
+    query: string;
+    database: PhmmerDatabase;
+    onProgress: (arg: string) => void;
+    onRid: (arg: string) => void;
+    signal?: AbortSignal;
+}): Promise<{
+    rows: PhmmerRow[];
+    /**
+     * the query, placed into the same columns. phmmer does not put the query in
+     * its own output, so this is derived — see buildQueryRow.
+     */
+    queryRow: string;
+    rid: string;
+}>;

package/dist/utils/phmmer.js

@@ -0,0 +1,120 @@
+import { StockholmMSA } from 'msa-parsers';
+import { fetchEbiResult, submitEbiJob, waitForEbiJob } from './ebiJobDispatcher';
+const TOOL = 'hmmer3_phmmer';
+/**
+ * Human-facing link to a job, shown while it runs and on error.
+ *
+ * The category has to be sss: jdispatcher serves its shell with a 200 for any
+ * category, so /pfa/ and /psa/ look fine to a fetch and render "Page Not Found"
+ * in a browser.
+ */
+export function phmmerResultUrl(jobId) {
+    return `https://www.ebi.ac.uk/jdispatcher/sss/${TOOL}/summary?jobId=${jobId}`;
+}
+/** EBI job ids are prefixed with the tool that made them */
+export function isPhmmerJobId(jobId) {
+    return jobId.startsWith(`${TOOL}-`);
+}
+/**
+ * phmmer aligns every hit to a profile built from the query, one match state
+ * per query residue, and marks those columns 'x' in #=GC RF. So the query's own
+ * row is exactly recoverable: walk RF, consume a query residue at each match
+ * column, gap everywhere else.
+ *
+ * This is the one piece of real logic here rather than a library call, and it
+ * is checked hard: if the match columns do not account for the query exactly,
+ * the columns and the query have drifted apart, and a query row that is off by
+ * even one residue would silently mis-map every column to the genome. Throwing
+ * is much better than drawing that.
+ */
+function buildQueryRow({ rf, query }) {
+    let consumed = 0;
+    const row = Array.from(rf, c => c === 'x' ? (query[consumed++] ?? '-') : '-').join('');
+    if (consumed !== query.length) {
+        throw new Error(`phmmer alignment has ${consumed} match columns for a query of ${query.length} residues, so the query row cannot be placed`);
+    }
+    return row;
+}
+/**
+ * '[subseq from] Albumin OS=Homo sapiens OX=9606 GN=ALB PE=1 SV=2' is what a
+ * UniProt target's #=GS DE looks like. Hits from the non-UniProt databases
+ * phmmer also offers (PDB, AlphaFold, MEROPS...) carry no OS=/OX= at all, so
+ * every field here is optional.
+ */
+function parseDescription(de) {
+    const text = (de ?? '').replace('[subseq from] ', '');
+    const sciname = /OS=(.*?)\s+(?:OX|GN|PE|SV)=/.exec(text)?.[1];
+    const ox = /OX=(\d+)/.exec(text)?.[1];
+    return {
+        sciname: sciname ?? 'unknown',
+        taxid: ox ? Number.parseInt(ox, 10) : undefined,
+        title: text.split(' OS=')[0] || undefined,
+    };
+}
+/**
+ * Target names look like 'sp|P02768|ALBU_HUMAN/1-609' for UniProt databases and
+ * like anything at all for the others, so an unrecognized name becomes its own
+ * accession rather than being dropped.
+ */
+function parseName(name) {
+    const slash = name.lastIndexOf('/');
+    const range = slash === -1 ? undefined : /^\d+-\d+$/.exec(name.slice(slash + 1))?.[0];
+    const bare = range === undefined ? name : name.slice(0, slash);
+    const parts = bare.split('|');
+    return parts.length === 3
+        ? { accession: parts[1], id: parts[2], range }
+        : { accession: bare, id: bare, range };
+}
+/**
+ * Exported for testing against a captured .sto — the annotation names (RF, the
+ * DE line's OS=/OX=) are the whole risk in this mapping, and nothing else in CI
+ * would notice if HMMER or EBI changed one.
+ */
+export function parsePhmmerAlignment({ stockholm, query, }) {
+    const { gc, gs, seqdata, seqname } = new StockholmMSA(stockholm, 0).getMSA();
+    const rf = gc.RF;
+    if (!rf) {
+        throw new Error('phmmer alignment has no #=GC RF line');
+    }
+    return {
+        queryRow: buildQueryRow({ rf, query }),
+        rows: seqname.map(name => ({
+            ...parseName(name),
+            ...parseDescription(gs.DE?.[name]?.[0]),
+            // insert columns come back lowercase with '.' for gaps; the MSA renderer
+            // looks colors up by the literal letter, so lowercase would draw
+            // uncolored. The insert columns stay visible as gaps in the query row.
+            aligned: (seqdata[name] ?? '').replaceAll('.', '-').toUpperCase(),
+        })),
+    };
+}
+export async function queryPhmmer({ query, database, onProgress, onRid, signal, }) {
+    onProgress('Submitting to EBI phmmer...');
+    const jobId = await submitEbiJob({
+        tool: TOOL,
+        params: {
+            database,
+            sequence: query,
+            // the alignment is the whole point of using phmmer here
+            alignView: 'true',
+        },
+        signal,
+    });
+    onRid(jobId);
+    await waitForEbiJob({
+        tool: TOOL,
+        jobId,
+        signal,
+        onCountdown: s => {
+            onProgress(`Re-checking phmmer status in... ${s}`);
+        },
+    });
+    const alignment = parsePhmmerAlignment({
+        stockholm: await fetchEbiResult({ tool: TOOL, jobId, type: 'sto', signal }),
+        query,
+    });
+    if (alignment.rows.length === 0) {
+        throw new Error('No hits found');
+    }
+    return { rid: jobId, ...alignment };
+}

package/dist/utils/poll.d.ts

@@ -3,9 +3,14 @@
  * false when still pending, and throws on failure. Between checks it counts down
  * `intervalSeconds`, calling `onCountdown` each second so the UI can show
  * progress.
+ *
+ * `signal` ends the loop with an AbortError: the countdown rejects on the tick
+ * it is aborted rather than at the end of the interval, so a cancelled launch
+ * stops within a second instead of running for the job's lifetime.
  */
-export declare function pollLoop({ check, intervalSeconds, onCountdown, }: {
+export declare function pollLoop({ check, intervalSeconds, onCountdown, signal, }: {
     check: () => Promise<boolean>;
     intervalSeconds: number;
     onCountdown: (secondsRemaining: number) => void;
+    signal?: AbortSignal;
 }): Promise<void>;

package/dist/utils/poll.js

@@ -4,16 +4,21 @@ import { timeout } from './fetch';
  * false when still pending, and throws on failure. Between checks it counts down
  * `intervalSeconds`, calling `onCountdown` each second so the UI can show
  * progress.
+ *
+ * `signal` ends the loop with an AbortError: the countdown rejects on the tick
+ * it is aborted rather than at the end of the interval, so a cancelled launch
+ * stops within a second instead of running for the job's lifetime.
  */
-export async function pollLoop({ check, intervalSeconds, onCountdown, }) {
+export async function pollLoop({ check, intervalSeconds, onCountdown, signal, }) {
     // eslint-disable-next-line @typescript-eslint/no-unnecessary-condition
     while (true) {
+        signal?.throwIfAborted();
         if (await check()) {
             return;
         }
         for (let i = intervalSeconds; i > 0; i--) {
             onCountdown(i);
-            await timeout(1000);
+            await timeout(1000, signal);
         }
     }
 }

package/dist/utils/taxonomyNames.d.ts

@@ -2,4 +2,4 @@ export interface TaxonomyInfo {
     sciname: string;
     commonName?: string;
 }
-export declare function fetchTaxonomyInfo(taxids: number[]): Promise<Map<number, TaxonomyInfo>>;
+export declare function fetchTaxonomyInfo(taxidsWithRepeats: number[]): Promise<Map<number, TaxonomyInfo>>;

package/dist/utils/taxonomyNames.js

@@ -25,7 +25,12 @@ async function saveTaxonomyCache(entries) {
     }
     await tx.done;
 }
-export async function fetchTaxonomyInfo(taxids) {
+export async function fetchTaxonomyInfo(taxidsWithRepeats) {
+    // callers pass one taxid per alignment row, and a BLAST hit list is several
+    // rows per species: 100 albumin hits are maybe 50 taxa, and asking as they
+    // came did 100 IndexedDB reads and sent eutils 100 ids for 50 answers. The
+    // result is keyed by taxid, so no caller can tell the difference
+    const taxids = [...new Set(taxidsWithRepeats)];
     const result = new Map();
     const uncachedTaxids = [];
     const cachedResults = await getCachedTaxonomies(taxids);

package/dist/version.d.ts

@@ -1 +1 @@
-export declare const version = "3.3.0";
+export declare const version = "3.4.1";

package/dist/version.js

@@ -1 +1 @@
-export const version = '3.3.0';
+export const version = '3.4.1';

package/package.json

@@ -1,5 +1,5 @@
 {
-  "version": "3.3.0",
+  "version": "3.4.1",
   "license": "MIT",
   "name": "jbrowse-plugin-msaview",
   "repository": {
@@ -17,8 +17,8 @@
   ],
   "dependencies": {
     "@emotion/styled": "^11.14.1",
-    "@gmod/bgzf-filehandle": "^6.2.0",
-    "g2p_mapper": "^2.1.5",
+    "@gmod/bgzf-filehandle": "^6.6.0",
+    "g2p_mapper": "^2.1.7",
     "idb": "^8.0.3"
   },
   "devDependencies": {
@@ -28,51 +28,57 @@
     "@jbrowse/core": "^4.3.0",
     "@jbrowse/mobx-state-tree": "^5.13.0",
     "@jbrowse/plugin-linear-genome-view": "^4.3.0",
-    "@mui/icons-material": "^9.2.0",
-    "@mui/material": "^9.2.0",
-    "@mui/system": "^9.2.0",
-    "@mui/x-data-grid": "^9.10.0",
+    "@mui/icons-material": "^9.4.0",
+    "@mui/material": "^9.4.0",
+    "@mui/system": "^9.4.0",
+    "@mui/x-data-grid": "^9.12.0",
     "@testing-library/dom": "^10.4.1",
-    "@testing-library/react": "^16.3.2",
-    "@types/node": "^26.1.1",
-    "@types/react": "^19.2.17",
-    "esbuild": "^0.28.1",
-    "eslint": "^10.7.0",
+    "@testing-library/react": "^16.3.3",
+    "@types/node": "^26.4.1",
+    "@types/react": "^19.2.18",
+    "esbuild": "^0.28.2",
+    "eslint": "^10.9.1",
     "eslint-plugin-import-x": "^4.17.1",
     "eslint-plugin-react": "^7.37.5",
     "eslint-plugin-react-hooks": "^7.1.1",
-    "eslint-plugin-unicorn": "^72.0.0",
+    "eslint-plugin-unicorn": "^74.0.0",
     "git-cliff": "^2.13.1",
     "jsdom": "^30.0.1",
     "mobx": "^6.16.1",
     "mobx-react": "^9.2.2",
-    "msa-parsers": "^6.0.0",
+    "msa-parsers": "^6.4.2",
+    "oxfmt": "^0.65.0",
+    "oxlint": "^1.81.0",
+    "oxlint-tsgolint": "^7.0.2001",
     "pixelmatch": "^7.2.0",
     "pngjs": "^7.0.0",
-    "prettier": "^3.9.6",
-    "pretty-bytes": "^7.1.1",
-    "puppeteer": "^25.3.0",
+    "pretty-bytes": "^7.1.2",
+    "puppeteer": "^25.9.0",
     "react": "^19.2.8",
     "react-dom": "^19.2.8",
-    "react-msaview": "^6.0.0",
+    "react-msaview": "^6.4.2",
     "rimraf": "^6.1.3",
     "rxjs": "^7.8.2",
     "serve": "^14.2.6",
     "tss-react": "^4.9.21",
-    "typescript": "^6.0.2",
-    "typescript-eslint": "^8.65.0",
-    "vitest": "^4.1.10"
+    "typescript": "^6.0.3",
+    "typescript-eslint": "^8.69.0",
+    "vitest": "^4.1.11"
   },
   "scripts": {
     "clean": "rimraf dist",
     "start": "node esbuild.mjs --watch",
-    "format": "pnpm prettier --write .",
+    "format": "oxfmt",
+    "check-format": "oxfmt --check",
     "build": "tsc && NODE_ENV=production node esbuild.mjs && cp distconfig.json dist/config.json",
     "prebuild": "pnpm clean",
-    "lint": "eslint src --report-unused-disable-directives --max-warnings 0",
+    "lint": "oxlint --type-aware --deny-warnings",
+    "lint:fast": "oxlint --deny-warnings",
+    "lint:eslint": "eslint src --report-unused-disable-directives --max-warnings 0",
     "check-text-source": "node scripts/check-text-source.mjs",
     "check-ebi-params": "node scripts/check-ebi-params.mjs",
     "check-mui-imports": "node scripts/check-mui-imports.mjs",
+    "check-host-dep-pins": "node scripts/check-host-dep-pins.mjs",
     "pretest": "rm -rf .test-jbrowse && npx @jbrowse/cli create .test-jbrowse --nightly",
     "test": "vitest run",
     "test:watch": "vitest",
@@ -82,7 +88,7 @@
     "test:version": "node scripts/test-versions.mjs run",
     "host-compat": "node scripts/host-compat-probe.mjs --bundle dist/jbrowse-plugin-msaview.umd.production.min.js",
     "check-ci": "node scripts/require-green-ci.mjs",
-    "preversion": "pnpm check-ci && pnpm lint && pnpm build && pnpm host-compat",
+    "preversion": "pnpm check-ci && pnpm check-host-dep-pins && pnpm lint && pnpm build && pnpm host-compat",
     "version": "node -e \"console.log('export const version = \\'' + require('./package.json').version + '\\'')\" > src/version.ts && git-cliff --tag v$npm_package_version --unreleased --prepend CHANGELOG.md && git add src/version.ts CHANGELOG.md",
     "postversion": "git push --follow-tags"
   }

package/src/AddHighlightModel/GenomeMouseoverHighlight.tsx

@@ -3,8 +3,8 @@ import React from 'react'
 import { getSession } from '@jbrowse/core/util'
 import { observer } from 'mobx-react'
 
-import { hasHoverPosition, useStyles } from './util'
 import { isMsaView } from '../MsaViewPanel/model'
+import { hasHoverPosition, useStyles } from './util'
 
 import type { LinearGenomeViewModel } from '@jbrowse/plugin-linear-genome-view'
 

package/src/AddHighlightModel/MsaToGenomeHighlight.tsx

@@ -3,8 +3,8 @@ import React from 'react'
 import { getSession } from '@jbrowse/core/util'
 import { observer } from 'mobx-react'
 
-import { hasHoverPosition, useStyles } from './util'
 import { isMsaView } from '../MsaViewPanel/model'
+import { hasHoverPosition, useStyles } from './util'
 
 import type { LinearGenomeViewModel } from '@jbrowse/plugin-linear-genome-view'
 

package/src/AddHighlightModel/index.tsx

@@ -2,8 +2,8 @@ import React from 'react'
 
 import { getSession } from '@jbrowse/core/util'
 
-import HighlightComponents from './HighlightComponents'
 import { isMsaView } from '../MsaViewPanel/model'
+import HighlightComponents from './HighlightComponents'
 
 import type PluginManager from '@jbrowse/core/PluginManager'
 import type { LinearGenomeViewModel } from '@jbrowse/plugin-linear-genome-view'