jbrowse-plugin-msaview 3.2.0 → 3.4.0

package/dist/utils/phmmer.js

@@ -0,0 +1,118 @@
+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, }) {
+    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',
+        },
+    });
+    onRid(jobId);
+    await waitForEbiJob({
+        tool: TOOL,
+        jobId,
+        onCountdown: s => {
+            onProgress(`Re-checking phmmer status in... ${s}`);
+        },
+    });
+    const alignment = parsePhmmerAlignment({
+        stockholm: await fetchEbiResult({ tool: TOOL, jobId, type: 'sto' }),
+        query,
+    });
+    if (alignment.rows.length === 0) {
+        throw new Error('No hits found');
+    }
+    return { rid: jobId, ...alignment };
+}

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 = "3.2.0";
+export declare const version = "3.4.0";

package/dist/version.js

@@ -1 +1 @@
-export const version = '3.2.0';
+export const version = '3.4.0';

package/package.json

@@ -1,5 +1,5 @@
 {
-  "version": "3.2.0",
+  "version": "3.4.0",
   "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": "^9.3.1",
+    "@mui/material": "^9.3.1",
+    "@mui/system": "^9.3.0",
+    "@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",
+    "@types/node": "^26.3.0",
+    "@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": "^73.0.0",
     "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": "^6.1.0",
+    "oxfmt": "^0.65.0",
+    "oxlint": "^1.80.0",
+    "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",
+    "puppeteer": "^25.8.0",
     "react": "^19.2.8",
     "react-dom": "^19.2.8",
-    "react-msaview": "^6.0.0",
+    "react-msaview": "^6.1.0",
     "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-eslint": "^8.68.0",
+    "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'

package/src/LaunchMsaView/components/BlastQuery/BlastAutomaticPanel.tsx

@@ -11,11 +11,6 @@ import {
 import { observer } from 'mobx-react'
 import { makeStyles } from 'tss-react/mui'
 
-import CachedBlastResults from './CachedBlastResults'
-import MsaAlgorithmSelect from './MsaAlgorithmSelect'
-import { blastLaunchView } from './blastLaunchView'
-import { blastDatabaseOptions, defaultBlastDatabase } from './consts'
-import { useCachedBlastResults } from './useCachedBlastResults'
 import TextField2 from '../../../components/TextField2'
 import {
   getBlastViewTitle,
@@ -26,14 +21,30 @@ import LaunchPanelContent from '../LaunchPanelContent'
 import SubmitCancelActions from '../SubmitCancelActions'
 import TranscriptSelector from '../TranscriptSelector'
 import { useTranscriptSelection } from '../useTranscriptSelection'
+import CachedBlastResults from './CachedBlastResults'
+import MsaAlgorithmSelect from './MsaAlgorithmSelect'
+import { blastLaunchView } from './blastLaunchView'
+import {
+  databaseOptionsFor,
+  defaultBlastDatabase,
+  defaultSearchFor,
+  searchPrograms,
+} from './consts'
+import { useCachedBlastResults } from './useCachedBlastResults'
 
-import type { BlastDatabase, MsaAlgorithm } from './consts'
+import type { MsaAlgorithm, SearchChoice, SearchProgram } from './consts'
 import type { AbstractTrackModel, Feature } from '@jbrowse/core/util'
 
 const useStyles = makeStyles()({
   selectField: {
     width: 150,
   },
+  // wider than the rest because the values are what the user came to read, and
+  // `uniprotkb_swissprot` is 19 characters — at 150 the field showed
+  // `uniprotkb_swis…`, which does not distinguish it from `uniprotkb_trembl`
+  databaseField: {
+    width: 230,
+  },
   cachedResultsAccordion: {
     marginTop: 20,
   },
@@ -56,10 +67,16 @@ const BlastAutomaticPanel = observer(function ({
   const { classes } = useStyles()
   const view = getLinearGenomeView(model)
   const [launchViewError, setLaunchViewError] = useState<unknown>()
-  const [selectedBlastDatabase, setSelectedBlastDatabase] =
-    useState<BlastDatabase>(defaultBlastDatabase)
+  // one piece of state, not two: a program and a database that program does not
+  // have is a 400 from EBI minutes after Submit, and holding them apart is what
+  // would let them drift into that
+  const [search, setSearch] = useState<SearchChoice>({
+    program: 'blastp',
+    database: defaultBlastDatabase,
+  })
   const [selectedMsaAlgorithm, setSelectedMsaAlgorithm] =
     useState<MsaAlgorithm>('clustalo')
+  const isPhmmer = search.program === 'phmmer'
 
   const geneIds = useMemo(() => getGeneIdentifiers(feature), [feature])
   const { results: cachedResults, error: cachedResultsError } =
@@ -74,37 +91,69 @@ const BlastAutomaticPanel = observer(function ({
         {children}
         <TextField2
           variant="outlined"
-          label="BLAST database"
+          label="Search program"
           className={classes.selectField}
           select
-          value={selectedBlastDatabase}
+          value={search.program}
           onChange={event => {
-            setSelectedBlastDatabase(event.target.value as BlastDatabase)
+            // the two services name their databases differently, so switching
+            // program replaces the database rather than keeping a name the new
+            // one has never heard of
+            setSearch(defaultSearchFor(event.target.value as SearchProgram))
           }}
         >
-          {blastDatabaseOptions.map(val => (
+          {searchPrograms.map(val => (
             <MenuItem value={val} key={val}>
               {val}
             </MenuItem>
           ))}
         </TextField2>
 
-        <MsaAlgorithmSelect
-          className={classes.selectField}
-          value={selectedMsaAlgorithm}
-          onChange={setSelectedMsaAlgorithm}
-        />
+        <TextField2
+          variant="outlined"
+          label="Database"
+          className={classes.databaseField}
+          select
+          value={search.database}
+          onChange={event => {
+            setSearch({
+              program: search.program,
+              database: event.target.value,
+            } as SearchChoice)
+          }}
+        >
+          {databaseOptionsFor(search.program).map(val => (
+            <MenuItem value={val} key={val}>
+              {val}
+            </MenuItem>
+          ))}
+        </TextField2>
+
+        {isPhmmer ? null : (
+          <MsaAlgorithmSelect
+            className={classes.selectField}
+            value={selectedMsaAlgorithm}
+            onChange={setSelectedMsaAlgorithm}
+          />
+        )}
 
         <TranscriptSelector feature={feature} {...transcriptSelection} />
 
         <Typography className={classes.infoText}>
-          This panel will automatically submit a blastp query to EBI, which
-          searches UniProtKB. Searches usually finish in under a minute, and
-          swissprot returns curated sequences that align more cleanly than the
-          many near-identical entries a TrEMBL search brings back. After
-          completion, all the hits will be run through a multiple sequence
-          alignment. Searching NCBI's nr needs the manual approach: NCBI no
-          longer lets a browser read responses from Blast.cgi.
+          {isPhmmer
+            ? `phmmer searches UniProtKB with a profile HMM built from the query,
+               so it aligns the hits as it finds them and that alignment is used
+               directly — nothing is realigned afterwards. The tree is then built
+               from it by neighbour-joining. A hit matching the query in more
+               than one place appears once per matched region.`
+            : `This panel will automatically submit a blastp query to EBI, which
+               searches UniProtKB. Searches usually finish in under a minute, and
+               swissprot returns curated sequences that align more cleanly than
+               the many near-identical entries a TrEMBL search brings back. After
+               completion, all the hits will be run through a multiple sequence
+               alignment.`}{' '}
+          Searching NCBI's nr needs the manual approach: NCBI no longer lets a
+          browser read responses from Blast.cgi.
         </Typography>
 
         {cachedResults.length > 0 ? (
@@ -133,12 +182,21 @@ const BlastAutomaticPanel = observer(function ({
                 feature: selectedTranscript,
                 view,
                 newViewTitle: getBlastViewTitle(feature, selectedTranscript),
-                blastParams: {
-                  blastDatabase: selectedBlastDatabase,
-                  msaAlgorithm: selectedMsaAlgorithm,
-                  selectedTranscript,
-                  proteinSequence,
-                },
+                blastParams:
+                  search.program === 'phmmer'
+                    ? {
+                        searchProgram: 'phmmer',
+                        blastDatabase: search.database,
+                        selectedTranscript,
+                        proteinSequence,
+                      }
+                    : {
+                        searchProgram: 'blastp',
+                        blastDatabase: search.database,
+                        msaAlgorithm: selectedMsaAlgorithm,
+                        selectedTranscript,
+                        proteinSequence,
+                      },
               })
               handleClose()
             }

package/src/LaunchMsaView/components/BlastQuery/BlastManualPanel.tsx

@@ -5,7 +5,6 @@ import { Alert, Typography } from '@mui/material'
 import { observer } from 'mobx-react'
 import { makeStyles } from 'tss-react/mui'
 
-import { BASE_BLAST_URL } from './consts'
 import ExternalLink from '../../../components/ExternalLink'
 import TextField2 from '../../../components/TextField2'
 import { useQueryRowName } from '../../useQueryRowName'
@@ -20,6 +19,7 @@ import QueryRowSelector from '../QueryRowSelector'
 import SubmitCancelActions from '../SubmitCancelActions'
 import TranscriptSelector from '../TranscriptSelector'
 import { useTranscriptSelection } from '../useTranscriptSelection'
+import { BASE_BLAST_URL } from './consts'
 
 import type { AbstractTrackModel, Feature } from '@jbrowse/core/util'
 

package/src/LaunchMsaView/components/BlastQuery/BlastPanel.tsx

@@ -4,15 +4,15 @@ import SettingsIcon from '@mui/icons-material/Settings'
 import { IconButton } from '@mui/material'
 import { makeStyles } from 'tss-react/mui'
 
-import BlastAutomaticPanel from './BlastAutomaticPanel'
-import BlastManualPanel from './BlastManualPanel'
-import BlastMethodSelector from './BlastMethodSelector'
-import BlastSettingsDialog from './BlastSettingsDialog'
 import {
   DEFAULT_EBI_EMAIL,
   EBI_EMAIL_STORAGE_KEY,
 } from '../../../utils/ebiJobDispatcher'
 import { useLocalStorage } from '../../../utils/useLocalStorage'
+import BlastAutomaticPanel from './BlastAutomaticPanel'
+import BlastManualPanel from './BlastManualPanel'
+import BlastMethodSelector from './BlastMethodSelector'
+import BlastSettingsDialog from './BlastSettingsDialog'
 
 import type { AbstractTrackModel, Feature } from '@jbrowse/core/util'
 

package/src/LaunchMsaView/components/BlastQuery/CachedBlastResults.test.ts

@@ -0,0 +1,50 @@
+import { expect, test } from 'vitest'
+
+import { describeSearch } from './CachedBlastResults'
+
+import type { CachedBlastResult } from '../../../utils/blastCache'
+
+const base: CachedBlastResult = {
+  id: 'k',
+  proteinSequence: 'MKV',
+  blastDatabase: 'uniprotkb_swissprot',
+  msa: '',
+  tree: '',
+  treeMetadata: '{}',
+  rid: 'r',
+  timestamp: 0,
+}
+
+function row(fields: Partial<CachedBlastResult>): CachedBlastResult {
+  return { ...base, ...fields }
+}
+
+test('a blastp row names its database, program and aligner', () => {
+  expect(
+    describeSearch(row({ searchProgram: 'blastp', msaAlgorithm: 'muscle' })),
+  ).toBe('uniprotkb_swissprot / blastp / muscle')
+})
+
+// phmmer aligns as it searches, so its rows carry no msaAlgorithm at all --
+// which read as "(undefined)" while this assumed one
+test('a phmmer row names no aligner, because none ran', () => {
+  expect(
+    describeSearch(
+      row({ blastDatabase: 'swissprot', searchProgram: 'phmmer' }),
+    ),
+  ).toBe('swissprot / phmmer')
+})
+
+test('a row cached before searchProgram existed reads as blastp', () => {
+  expect(describeSearch(row({ msaAlgorithm: 'clustalo' }))).toBe(
+    'uniprotkb_swissprot / blastp / clustalo',
+  )
+})
+
+test('a row from the NCBI era keeps the program it recorded', () => {
+  expect(
+    describeSearch(
+      row({ blastProgram: 'quick-blastp', msaAlgorithm: 'clustalo' }),
+    ),
+  ).toBe('uniprotkb_swissprot / quick-blastp / clustalo')
+})

package/src/LaunchMsaView/components/BlastQuery/CachedBlastResults.tsx

@@ -14,14 +14,14 @@ import {
 import { observer } from 'mobx-react'
 import { makeStyles } from 'tss-react/mui'
 
-import { blastLaunchViewFromCache } from './blastLaunchView'
-import { useCachedBlastResults } from './useCachedBlastResults'
 import {
   featureMatchesId,
   getGeneIdentifiers,
   getLinearGenomeView,
   getSortedTranscriptFeatures,
 } from '../../util'
+import { blastLaunchViewFromCache } from './blastLaunchView'
+import { useCachedBlastResults } from './useCachedBlastResults'
 
 import type { CachedBlastResult } from '../../../utils/blastCache'
 import type { AbstractTrackModel, Feature } from '@jbrowse/core/util'
@@ -51,6 +51,26 @@ function getResultDisplayName(result: CachedBlastResult): string {
     : (result.geneId ?? result.transcriptId ?? 'Unknown')
 }
 
+/**
+ * How the row was produced: `uniprotkb_swissprot / blastp / clustalo`, or
+ * `swissprot / phmmer` for a row phmmer aligned as it searched and that
+ * therefore ran no aligner. Each part is dropped when absent rather than
+ * printed empty — `msaAlgorithm` became optional when phmmer arrived, and a
+ * phmmer row read `(undefined)` until this stopped assuming one.
+ *
+ * `blastProgram` is the older field, written only while the plugin still
+ * queried NCBI directly and blastp/quick-blastp was a real choice.
+ */
+export function describeSearch(result: CachedBlastResult) {
+  return [
+    result.blastDatabase,
+    result.searchProgram ?? result.blastProgram ?? 'blastp',
+    result.msaAlgorithm,
+  ]
+    .filter(Boolean)
+    .join(' / ')
+}
+
 const CachedBlastResults = observer(function ({
   model,
   handleClose,
@@ -149,7 +169,7 @@ const CachedBlastResults = observer(function ({
               }}
             >
               <ListItemText
-                primary={`${getResultDisplayName(result)} - ${result.blastDatabase}${result.blastProgram ? `/${result.blastProgram}` : ''} (${result.msaAlgorithm})`}
+                primary={`${getResultDisplayName(result)} - ${describeSearch(result)}`}
                 secondary={`${new Date(result.timestamp).toLocaleString()} - Seq: ${result.proteinSequence.slice(0, 30)}...`}
               />
             </ListItemButton>

package/src/LaunchMsaView/components/BlastQuery/MsaAlgorithmSelect.tsx

@@ -2,8 +2,8 @@ import React from 'react'
 
 import { MenuItem } from '@mui/material'
 
-import { msaAlgorithms } from './consts'
 import TextField2 from '../../../components/TextField2'
+import { msaAlgorithms } from './consts'
 
 import type { MsaAlgorithm } from './consts'
 

package/src/LaunchMsaView/components/BlastQuery/consts.ts

@@ -27,3 +27,43 @@ export type BlastDatabase = (typeof blastDatabaseOptions)[number]
 // curated, so it returns roughly one good sequence per species rather than the
 // many near-identical TrEMBL entries an alignment reads poorly
 export const defaultBlastDatabase: BlastDatabase = 'uniprotkb_swissprot'
+
+export const searchPrograms = ['blastp', 'phmmer'] as const
+export type SearchProgram = (typeof searchPrograms)[number]
+
+/**
+ * phmmer offers PDB, AlphaFold, Ensembl Genomes, MEROPS and ChEMBL too, but
+ * targets outside UniProt carry no OS=/OX= in their description, so those rows
+ * would lose their species and common name. Only the databases that label their
+ * hits are offered.
+ */
+export const phmmerDatabaseOptions = [
+  'swissprot',
+  'uniprotkb',
+  'uniprotrefprot',
+] as const
+export type PhmmerDatabase = (typeof phmmerDatabaseOptions)[number]
+
+export const defaultPhmmerDatabase: PhmmerDatabase = 'swissprot'
+
+/**
+ * A program together with a database that program actually has.
+ *
+ * The pair travels as one value because neither service knows the other's
+ * database names — `swissprot` is a phmmer database and `uniprotkb_swissprot` a
+ * blastp one — so a program held apart from its database can drift into a
+ * combination EBI answers with a 400, minutes after the user pressed Submit.
+ */
+export type SearchChoice =
+  | { program: 'blastp'; database: BlastDatabase }
+  | { program: 'phmmer'; database: PhmmerDatabase }
+
+export function defaultSearchFor(program: SearchProgram): SearchChoice {
+  return program === 'phmmer'
+    ? { program, database: defaultPhmmerDatabase }
+    : { program, database: defaultBlastDatabase }
+}
+
+export function databaseOptionsFor(program: SearchProgram) {
+  return program === 'phmmer' ? phmmerDatabaseOptions : blastDatabaseOptions
+}