jbrowse-plugin-msaview 5.16.0 → 5.16.2

package/dist/BgzipFastaMsaAdapter/BgzipFastaMsaAdapter.d.ts

@@ -1,12 +1,30 @@
 import { BaseAdapter } from '@jbrowse/core/data_adapters/BaseAdapter';
+import type { MsaIndexEntry } from '../utils/msaIndex';
 import type { BaseFeatureDataAdapter } from '@jbrowse/core/data_adapters/BaseAdapter';
+import type { FileLocation } from '@jbrowse/core/util';
 export default class BgzipFastaMsaAdapter extends BaseAdapter {
     configureP: Promise<BaseFeatureDataAdapter> | undefined;
     msaRowsP: Promise<Map<string, string[]>> | undefined;
+    msaIndexP: Promise<Map<string, MsaIndexEntry> | undefined> | undefined;
     configurePre(): Promise<BaseFeatureDataAdapter<import("@jbrowse/core/configuration").AnyConfigurationModel>>;
     configure(): Promise<BaseFeatureDataAdapter<import("@jbrowse/core/configuration").AnyConfigurationModel>>;
+    readMsaIndex(location: FileLocation): Promise<Map<string, {
+        name: string;
+        offset: number;
+        length: number;
+    }>>;
+    loadMsaIndex(): Promise<Map<string, {
+        name: string;
+        offset: number;
+        length: number;
+    }> | undefined>;
+    getMsaIndex(): Promise<Map<string, MsaIndexEntry> | undefined>;
     groupRows(): Promise<Map<string, string[]>>;
     getMSARows(): Promise<Map<string, string[]>>;
     getMSAList(): Promise<string[]>;
-    getMSA(id: string): Promise<import("@jbrowse/core/util").Feature[]>;
+    readIndexedMSA(index: Map<string, MsaIndexEntry>, id: string): Promise<string>;
+    readRowsMSA(id: string): Promise<string>;
+    getMSA(id: string): Promise<{
+        msa: string;
+    }>;
 }

package/dist/BgzipFastaMsaAdapter/BgzipFastaMsaAdapter.js

@@ -1,6 +1,8 @@
 import { readConfObject } from '@jbrowse/core/configuration';
 import { BaseAdapter } from '@jbrowse/core/data_adapters/BaseAdapter';
+import { openLocation } from '@jbrowse/core/util/io';
 import { firstValueFrom, toArray } from 'rxjs';
+import { parseMsaIndex, readMsaBlock } from '../utils/msaIndex';
 export default class BgzipFastaMsaAdapter extends BaseAdapter {
     async configurePre() {
         const getSubAdapter = this.getSubAdapter;
@@ -22,6 +24,29 @@ export default class BgzipFastaMsaAdapter extends BaseAdapter {
         });
         return this.configureP;
     }
+    async readMsaIndex(location) {
+        const text = await openLocation(location, this.pluginManager).readFile('utf8');
+        return new Map(parseMsaIndex(text).map(e => [e.name, e]));
+    }
+    // an `.idx` beside the FASTA is used when present, as the host finds the
+    // `.fai` and `.gzi`; only one named in the config is required to exist
+    async loadMsaIndex() {
+        const configured = this.getConf('msaIndexLocation');
+        const adapter = await this.configure();
+        const beside = withSuffix(adapter.getConf('fastaLocation'), '.idx');
+        return 'uri' in configured && !configured.uri
+            ? beside
+                ? this.readMsaIndex(beside).catch(() => undefined)
+                : undefined
+            : this.readMsaIndex(configured);
+    }
+    getMsaIndex() {
+        this.msaIndexP ??= this.loadMsaIndex().catch((e) => {
+            this.msaIndexP = undefined;
+            throw e;
+        });
+        return this.msaIndexP;
+    }
     async groupRows() {
         const adapter = await this.configure();
         const separator = new RegExp(this.getConf('msaRegex'));
@@ -46,12 +71,23 @@ export default class BgzipFastaMsaAdapter extends BaseAdapter {
         return this.msaRowsP;
     }
     async getMSAList() {
-        return [...(await this.getMSARows()).keys()];
+        return [...((await this.getMsaIndex()) ?? (await this.getMSARows())).keys()];
     }
-    async getMSA(id) {
+    async readIndexedMSA(index, id) {
+        const adapter = await this.configure();
+        const entry = index.get(id);
+        return entry
+            ? readMsaBlock({
+                filehandle: openLocation(adapter.getConf('fastaLocation'), this.pluginManager),
+                gziFilehandle: openLocation(adapter.getConf('gziLocation'), this.pluginManager),
+                entry,
+            })
+            : '';
+    }
+    async readRowsMSA(id) {
         const adapter = await this.configure();
         const rows = (await this.getMSARows()).get(id) ?? [];
-        return firstValueFrom(adapter
+        const features = await firstValueFrom(adapter
             .getFeaturesInMultipleRegions(rows.map(refName => ({
             refName,
             start: 0,
@@ -59,5 +95,21 @@ export default class BgzipFastaMsaAdapter extends BaseAdapter {
             assemblyName: '',
         })))
             .pipe(toArray()));
+        return features.map(f => `>${f.get('refName')}\n${f.get('seq')}`).join('\n');
     }
+    async getMSA(id) {
+        const index = await this.getMsaIndex();
+        return {
+            msa: await (index
+                ? this.readIndexedMSA(index, id)
+                : this.readRowsMSA(id)),
+        };
+    }
+}
+function withSuffix(location, suffix) {
+    return 'uri' in location
+        ? { ...location, uri: `${location.uri}${suffix}` }
+        : 'localPath' in location
+            ? { ...location, localPath: `${location.localPath}${suffix}` }
+            : undefined;
 }

package/dist/BgzipFastaMsaAdapter/configSchema.d.ts

@@ -5,10 +5,20 @@ export default function configSchemaF(pluginManager: PluginManager): import("@jb
     } & {
         defaultValue: string;
     };
+    msaIndexLocation: {
+        type: "fileLocation";
+    } & {
+        defaultValue: unknown;
+    };
 } | {
     readonly msaRegex: {
         type: "string";
     } & {
         defaultValue: string;
     };
+    readonly msaIndexLocation: {
+        type: "fileLocation";
+    } & {
+        defaultValue: unknown;
+    };
 }, import("@jbrowse/core/configuration/configurationSchema").ConfigurationSchemaOptions<any, undefined, undefined, import("@jbrowse/core/configuration").ConfigurationSchemaRequirement<Partial<Record<string, string[]>>, string>, any>>;

package/dist/BgzipFastaMsaAdapter/configSchema.js

@@ -6,6 +6,11 @@ export default function configSchemaF(pluginManager) {
             type: 'string',
             defaultValue: '_',
         },
+        msaIndexLocation: {
+            type: 'fileLocation',
+            description: '`.idx` of name, offset and length per alignment, which lists the set and reads each alignment as one block in place of the per-row .fai. Found beside the FASTA when unset',
+            defaultValue: { uri: '', locationType: 'UriLocation' },
+        },
     }, {
         baseConfiguration: base.configSchema,
     });

package/dist/LaunchMsaView/components/PreLoadedMSA/PreLoadedMSADataPanel.js

@@ -7,7 +7,7 @@ import { makeStyles } from 'tss-react/mui';
 import TextField2 from '../../../components/TextField2';
 import { useFetch } from '../../../utils/useFetch';
 import { resolveQueryRowName, useQueryRowName } from '../../useQueryRowName';
-import { featureMatchesId, getGeneDisplayName, getLinearGenomeView, getTranscriptDisplayName, } from '../../util';
+import { findDatasetId, getGeneDisplayName, getLinearGenomeView, getTranscriptDisplayName, } from '../../util';
 import LaunchPanelContent from '../LaunchPanelContent';
 import QueryRowSelector from '../QueryRowSelector';
 import SequenceStatusMessage from '../SequenceStatus';
@@ -28,43 +28,53 @@ const PreLoadedMSA = observer(function ({ model, feature, handleClose, preferred
     const { classes } = useStyles();
     const { pluginManager } = getEnv(model);
     const { assemblyNames } = view;
+    const assemblyName = assemblyNames[0];
     const { launchError, submit } = useLaunchSubmit(handleClose);
     const datasets = readMsaDatasets(session.jbrowse);
     const [selectedDatasetId, setSelectedDatasetId] = useState(datasets?.[0]?.datasetId);
     const selectedDataset = datasets?.find(d => d.datasetId === selectedDatasetId);
-    const { data: msaList, isLoading: msaListLoading, error: msaListFetchError, } = useFetch(selectedDataset ? `${selectedDataset.datasetId}-msa-list` : null, () => fetchMSAList({ config: selectedDataset.adapter, pluginManager }));
+    const { data: msaList, isLoading: msaListLoading, error: msaListFetchError, } = useFetch(selectedDataset
+        ? `${selectedDataset.datasetId}-${assemblyName}-msa-list`
+        : null, () => fetchMSAList({
+        config: selectedDataset.adapter,
+        pluginManager,
+        assemblyName,
+    }));
+    const msaIds = useMemo(() => msaList && new Set(msaList), [msaList]);
     const transcriptSelection = useTranscriptSelection({
         feature,
         view,
-        validIds: msaList,
+        hasData: msaIds
+            ? transcript => !!findDatasetId(transcript, feature, msaIds)
+            : undefined,
         preferredTranscriptId,
     });
     const { selectedTranscript, proteinSequence, sequenceStatus } = transcriptSelection;
     // the dataset's name for this transcript, which a track is free not to use
     // as the feature's own id
-    const msaId = selectedTranscript
-        ? msaList?.find(id => featureMatchesId(selectedTranscript, id))
+    const msaId = selectedTranscript && msaIds
+        ? findDatasetId(selectedTranscript, feature, msaIds)
         : undefined;
     const { data: msaData, isLoading: msaDataLoading, error: msaDataFetchError, } = useFetch(msaId && selectedDataset
-        ? `${selectedDataset.datasetId}-${msaId}-msa`
+        ? `${selectedDataset.datasetId}-${assemblyName}-${msaId}-msa`
         : null, () => fetchMSA({
         msaId: msaId,
+        assemblyName,
         config: selectedDataset.adapter,
         pluginManager,
     }));
-    const msaText = useMemo(() => msaData?.map(r => `>${r.get('refName')}\n${r.get('seq')}`).join('\n') ??
-        '', [msaData]);
+    const msaText = msaData?.msa ?? '';
     // The dataset's row for this transcript used to be assumed -- the launch
     // named `<transcriptId>_<assembly>` and hoped the file agreed. Nothing
     // checked, and a name the alignment does not carry fails silently: the view
     // opens, renders, and never navigates. The row is found by sequence instead,
     // the same way the Manual tab finds it.
-    const queryRow = useQueryRowName(msaText, proteinSequence);
+    const queryRow = useQueryRowName(msaText, proteinSequence, msaData?.queryRowName);
     // The name this panel used to launch with unconditionally. A dataset built to
     // that convention does carry the row, so it is worth falling back to when the
     // residues do not match closely enough to find it -- but only when the
     // alignment really has it, which is the check the old code never made.
-    const querySeqName = resolveQueryRowName(queryRow, `${msaId ?? ''}_${assemblyNames[0] ?? ''}`);
+    const querySeqName = resolveQueryRowName(queryRow, `${msaId ?? ''}_${assemblyName ?? ''}`);
     const e = msaListFetchError ??
         msaDataFetchError ??
         transcriptSelection.error ??
@@ -81,11 +91,11 @@ const PreLoadedMSA = observer(function ({ model, feature, handleClose, preferred
                     React.createElement(SanitizedHTML, { html: selectedDataset.description }),
                     React.createElement(TranscriptSelector, { feature: feature, ...transcriptSelection }),
                     msaText ? (React.createElement(QueryRowSelector, { ...queryRow, querySeqName: querySeqName })) : null)) : null)) : null),
-        React.createElement(SubmitCancelActions, { model: model, hint: msaDataLoading ? (React.createElement(LoadingEllipses, { message: "Loading alignment" })) : !!msaData?.length && !querySeqName ? (React.createElement(Typography, { color: "textSecondary", variant: "body2" }, "no row matches this transcript")) : (React.createElement(SequenceStatusMessage, { status: sequenceStatus })), 
+        React.createElement(SubmitCancelActions, { model: model, hint: msaDataLoading ? (React.createElement(LoadingEllipses, { message: "Loading alignment" })) : !!msaText && !querySeqName ? (React.createElement(Typography, { color: "textSecondary", variant: "body2" }, "no row matches this transcript")) : (React.createElement(SequenceStatusMessage, { status: sequenceStatus })), 
             // launching without a query row opens a view that renders and then
             // never navigates, which reads as a broken feature rather than a
             // dataset that does not cover this gene
-            submitDisabled: !selectedTranscript || !msaData?.length || !querySeqName, onSubmit: placement => {
+            submitDisabled: !selectedTranscript || !msaText || !querySeqName, onSubmit: placement => {
                 if (selectedTranscript && msaText) {
                     submit(() => launchConnectedView({
                         view,
@@ -100,7 +110,7 @@ const PreLoadedMSA = observer(function ({ model, feature, handleClose, preferred
                         colorSchemeName: 'percent_identity_dynamic',
                         querySeqName,
                         querySeqOffset: queryRow.querySeqOffset,
-                        data: { msa: msaText },
+                        data: { msa: msaText, tree: msaData?.tree },
                     }));
                 }
             }, onCancel: handleClose })));

package/dist/LaunchMsaView/components/PreLoadedMSA/fetchMSAData.d.ts

@@ -1,12 +1,19 @@
 import type PluginManager from '@jbrowse/core/PluginManager';
 import type { AnyConfigurationModel } from '@jbrowse/core/configuration';
-import type { Feature } from '@jbrowse/core/util';
-export declare function fetchMSAList({ config, pluginManager, }: {
+export interface DatasetMsa {
+    msa: string;
+    tree?: string;
+    /** the row the dataset names as the protein's own, preferred over a row that matches it equally */
+    queryRowName?: string;
+}
+export declare function fetchMSAList({ config, pluginManager, assemblyName, }: {
     config: AnyConfigurationModel;
     pluginManager: PluginManager;
+    assemblyName?: string;
 }): Promise<string[]>;
-export declare function fetchMSA({ config, pluginManager, msaId, }: {
+export declare function fetchMSA({ config, pluginManager, msaId, assemblyName, }: {
     config: AnyConfigurationModel;
     pluginManager: PluginManager;
     msaId: string;
-}): Promise<Feature[]>;
+    assemblyName?: string;
+}): Promise<DatasetMsa>;

package/dist/LaunchMsaView/components/PreLoadedMSA/fetchMSAData.js

@@ -3,11 +3,11 @@ async function getMsaAdapter(pluginManager, config) {
     const result = await getAdapter(pluginManager, 'msa', config);
     return result.dataAdapter;
 }
-export async function fetchMSAList({ config, pluginManager, }) {
+export async function fetchMSAList({ config, pluginManager, assemblyName, }) {
     const adapter = await getMsaAdapter(pluginManager, config);
-    return adapter.getMSAList();
+    return adapter.getMSAList(assemblyName);
 }
-export async function fetchMSA({ config, pluginManager, msaId, }) {
+export async function fetchMSA({ config, pluginManager, msaId, assemblyName, }) {
     const adapter = await getMsaAdapter(pluginManager, config);
-    return adapter.getMSA(msaId);
+    return adapter.getMSA(msaId, assemblyName);
 }

package/dist/LaunchMsaView/components/TranscriptSelector.d.ts

@@ -1,7 +1,7 @@
 import React from 'react';
 import type { SequenceStatus } from './SequenceStatus';
 import type { Feature } from '@jbrowse/core/util';
-declare const TranscriptSelector: ({ feature, options, selectedId, selectedTranscript, setSelectedId, proteinSequence, sequenceStatus, validIds, }: {
+declare const TranscriptSelector: ({ feature, options, selectedId, selectedTranscript, setSelectedId, proteinSequence, sequenceStatus, hasData, }: {
     feature: Feature;
     options: Feature[];
     selectedId: string;
@@ -9,6 +9,6 @@ declare const TranscriptSelector: ({ feature, options, selectedId, selectedTrans
     setSelectedId: (transcriptId: string) => void;
     proteinSequence: string | undefined;
     sequenceStatus?: SequenceStatus;
-    validIds?: string[];
+    hasData?: (transcript: Feature) => boolean;
 }) => React.JSX.Element;
 export default TranscriptSelector;

package/dist/LaunchMsaView/components/TranscriptSelector.js

@@ -3,7 +3,7 @@ import { Button, MenuItem, TextField } from '@mui/material';
 import { observer } from 'mobx-react';
 import { makeStyles } from 'tss-react/mui';
 import ReadOnlyTextField2 from '../../components/ReadOnlyTextField2';
-import { featureMatchesId, getGeneDisplayName, getId, getTranscriptDisplayName, getTranscriptLength, } from '../util';
+import { getGeneDisplayName, getId, getTranscriptDisplayName, getTranscriptLength, } from '../util';
 const useStyles = makeStyles()({
     flex: {
         display: 'flex',
@@ -16,7 +16,7 @@ const useStyles = makeStyles()({
         marginLeft: 20,
     },
 });
-const TranscriptSelector = observer(function TranscriptSelector({ feature, options, selectedId, selectedTranscript, setSelectedId, proteinSequence, sequenceStatus = 'ready', validIds, }) {
+const TranscriptSelector = observer(function TranscriptSelector({ feature, options, selectedId, selectedTranscript, setSelectedId, proteinSequence, sequenceStatus = 'ready', hasData, }) {
     const { classes } = useStyles();
     const [showSequence, setShowSequence] = useState(false);
     return (React.createElement(React.Fragment, null,
@@ -30,9 +30,7 @@ const TranscriptSelector = observer(function TranscriptSelector({ feature, optio
                 helperText: "the query row, so the alignment stays linked to the genome view", select: true, className: classes.minWidth, value: selectedId, onChange: event => {
                     setSelectedId(event.target.value);
                 } }, options.map(val => {
-                const inSet = validIds
-                    ? validIds.some(id => featureMatchesId(val, id))
-                    : true;
+                const inSet = hasData ? hasData(val) : true;
                 const { len, mod } = getTranscriptLength(val);
                 return (React.createElement(MenuItem, { value: getId(val), key: val.id(), disabled: !inSet },
                     getTranscriptDisplayName(val),
@@ -41,7 +39,7 @@ const TranscriptSelector = observer(function TranscriptSelector({ feature, optio
                     " aa)",
                     ' ',
                     mod ? ` (possible fragment)` : '',
-                    validIds ? (inSet ? ' (has data)' : ' (no data)') : ''));
+                    hasData ? (inSet ? ' (has data)' : ' (no data)') : ''));
             })),
             React.createElement("div", { className: classes.centered },
                 React.createElement(Button, { variant: "contained", color: "primary", onClick: () => {

package/dist/LaunchMsaView/components/pickSelectedId.test.d.ts

@@ -0,0 +1 @@
+export {};

package/dist/LaunchMsaView/components/pickSelectedId.test.js

@@ -0,0 +1,17 @@
+import { SimpleFeature } from '@jbrowse/core/util';
+import { expect, test } from 'vitest';
+import { pickSelectedId } from './useTranscriptSelection';
+const transcript = (id) => new SimpleFeature({ uniqueId: id, refName: '2L', start: 0, end: 10 });
+const options = ['t1', 't2', 't3'].map(transcript);
+test('with no hasData the selection always stands', () => {
+    expect(pickSelectedId('t2', options, undefined)).toBe('t2');
+});
+test('a selection with data stands', () => {
+    expect(pickSelectedId('t2', options, () => true)).toBe('t2');
+});
+test('a selection without data moves to the first option that has some', () => {
+    expect(pickSelectedId('t1', options, t => t.id() !== 't1')).toBe('t2');
+});
+test('with no option holding data the selection stands', () => {
+    expect(pickSelectedId('t3', options, () => false)).toBe('t3');
+});

package/dist/LaunchMsaView/components/useTranscriptSelection.d.ts

@@ -1,11 +1,13 @@
 import type { SequenceStatus } from './SequenceStatus';
 import type { Feature } from '@jbrowse/core/util';
-export declare function useTranscriptSelection({ feature, view, validIds, preferredTranscriptId, }: {
+export declare function pickSelectedId(currentId: string, options: Feature[], hasData: ((transcript: Feature) => boolean) | undefined): string;
+export declare function useTranscriptSelection({ feature, view, hasData, preferredTranscriptId, }: {
     feature: Feature;
     view: {
         assemblyNames?: string[];
     } | undefined;
-    validIds?: string[];
+    /** which transcripts the panel can offer, when it can only offer some */
+    hasData?: (transcript: Feature) => boolean;
     /**
      * the isoform the user right-clicked, when the dialog opened on its gene.
      * Ignored when this gene does not carry it, so a stale id falls back to the
@@ -21,5 +23,5 @@ export declare function useTranscriptSelection({ feature, view, validIds, prefer
     error: unknown;
     isLoading: boolean;
     sequenceStatus: SequenceStatus;
-    validIds: string[] | undefined;
+    hasData: ((transcript: Feature) => boolean) | undefined;
 };

package/dist/LaunchMsaView/components/useTranscriptSelection.js

@@ -1,19 +1,18 @@
 import { useMemo, useState } from 'react';
 import { featureMatchesId, getId, getSortedTranscriptFeatures } from '../util';
 import { useFeatureSequence } from './useFeatureSequence';
-// Keep the current selection if it's valid for the given validIds, otherwise
-// fall back to the first valid option (or the current id if none qualify). With
-// no validIds constraint the current selection always stands.
-function pickSelectedId(currentId, options, validIds) {
-    if (!validIds?.length) {
+// Keep the current selection if it has data, otherwise fall back to the first
+// option that does (or the current id if none do). With no hasData constraint
+// the current selection always stands.
+export function pickSelectedId(currentId, options, hasData) {
+    if (!hasData) {
         return currentId;
     }
-    const isValid = (opt) => validIds.some(id => featureMatchesId(opt, id));
     const current = options.find(opt => getId(opt) === currentId);
-    if (current && isValid(current)) {
+    if (current && hasData(current)) {
         return currentId;
     }
-    const firstValid = options.find(isValid);
+    const firstValid = options.find(hasData);
     return firstValid ? getId(firstValid) : currentId;
 }
 /**
@@ -31,11 +30,11 @@ function sequenceStatus({ error, isLoading, proteinSequence, }) {
     }
     return proteinSequence ? 'ready' : 'missing';
 }
-export function useTranscriptSelection({ feature, view, validIds, preferredTranscriptId, }) {
+export function useTranscriptSelection({ feature, view, hasData, preferredTranscriptId, }) {
     const options = useMemo(() => getSortedTranscriptFeatures(feature), [feature]);
     const [selectedId, setSelectedId] = useState(() => getId(options.find(opt => featureMatchesId(opt, preferredTranscriptId ?? '')) ??
         options[0]));
-    const validatedSelectedId = pickSelectedId(selectedId, options, validIds);
+    const validatedSelectedId = pickSelectedId(selectedId, options, hasData);
     const selectedTranscript = options.find(val => getId(val) === validatedSelectedId);
     const { proteinSequence, error, isLoading } = useFeatureSequence({
         view,
@@ -50,6 +49,6 @@ export function useTranscriptSelection({ feature, view, validIds, preferredTrans
         error,
         isLoading,
         sequenceStatus: sequenceStatus({ error, isLoading, proteinSequence }),
-        validIds,
+        hasData,
     };
 }

package/dist/LaunchMsaView/detectQueryRow.d.ts

@@ -44,4 +44,4 @@ export interface MsaQueryRow {
  * wanted on every keystroke in the paste box, and asking separately parsed a
  * few-hundred-row alignment twice per character.
  */
-export declare function findQueryRow(msaText: string, proteinSequence: string): MsaQueryRow;
+export declare function findQueryRow(msaText: string, proteinSequence: string, preferredName?: string): MsaQueryRow;

package/dist/LaunchMsaView/detectQueryRow.js

@@ -41,7 +41,7 @@ const PARTIAL_COVERAGE_FLOOR = 0.5;
  * wanted on every keystroke in the paste box, and asking separately parsed a
  * few-hundred-row alignment twice per character.
  */
-export function findQueryRow(msaText, proteinSequence) {
+export function findQueryRow(msaText, proteinSequence, preferredName) {
     if (!msaText.trim()) {
         return { names: [] };
     }
@@ -56,42 +56,54 @@ export function findQueryRow(msaText, proteinSequence) {
     }
     const names = parsed.getNames();
     const query = normalize(proteinSequence);
-    return { names, match: query ? bestMatch(parsed, names, query) : undefined };
+    return {
+        names,
+        match: query ? bestMatch(parsed, names, query, preferredName) : undefined,
+    };
 }
-function bestMatch(parsed, names, query) {
-    const candidates = [];
-    for (const name of names) {
-        const row = normalize(getUngappedSequence(parsed.getRow(name)));
-        if (!row) {
-            continue;
-        }
-        if (row === query) {
-            // nothing beats an exact match, and a second one would be a duplicate row
-            return { name, quality: 'exact', identity: 1, offset: 0 };
-        }
+function scoreRow(parsed, name, query) {
+    const row = normalize(getUngappedSequence(parsed.getRow(name)));
+    if (!row) {
+        return undefined;
+    }
+    else if (row === query) {
+        return { name, quality: 'exact', identity: 1, offset: 0 };
+    }
+    else {
         // BLAST reports the aligned region, so the row is often the query trimmed
         // at one or both ends rather than the whole protein
         const contained = query.indexOf(row);
         const containing = row.indexOf(query);
         if (contained !== -1 || containing !== -1) {
             const coverage = Math.min(row.length, query.length) / Math.max(row.length, query.length);
-            if (coverage >= PARTIAL_COVERAGE_FLOOR) {
-                candidates.push({
+            return coverage >= PARTIAL_COVERAGE_FLOOR
+                ? {
                     name,
                     quality: 'partial',
                     identity: coverage,
                     offset: contained === -1 ? -containing : contained,
-                });
-            }
-            continue;
+                }
+                : undefined;
         }
-        const identity = identityOverOverlap(row, query);
-        if (identity >= SIMILARITY_FLOOR) {
-            candidates.push({ name, quality: 'similar', identity, offset: 0 });
+        else {
+            const identity = identityOverOverlap(row, query);
+            return identity >= SIMILARITY_FLOOR
+                ? { name, quality: 'similar', identity, offset: 0 }
+                : undefined;
         }
     }
-    // an exact match returns above, so only these two can be here
-    const order = ['partial', 'similar'];
-    return candidates.sort((a, b) => order.indexOf(a.quality) - order.indexOf(b.quality) ||
-        b.identity - a.identity)[0];
+}
+const qualityOrder = ['exact', 'partial', 'similar'];
+/**
+ * The row matching the protein best. A paralog with an identical protein
+ * matches exactly as well as the query does, so a row the dataset itself names
+ * as the protein's own wins any tie.
+ */
+function bestMatch(parsed, names, query, preferredName) {
+    return names
+        .map(name => scoreRow(parsed, name, query))
+        .filter(match => match !== undefined)
+        .sort((a, b) => qualityOrder.indexOf(a.quality) - qualityOrder.indexOf(b.quality) ||
+        b.identity - a.identity ||
+        Number(b.name === preferredName) - Number(a.name === preferredName))[0];
 }

package/dist/LaunchMsaView/detectQueryRow.test.js

@@ -93,3 +93,12 @@ describe('findQueryRow row names', () => {
         expect(findQueryRow('', protein).names).toEqual([]);
     });
 });
+test("an identical paralog loses the tie to the row the dataset names as the protein's own", () => {
+    const msa = '>sp_H4copy\nMSGRGKGG\n>sp_His4r\nMSGRGKGG\n>sp_other\nMSGRGKAA\n';
+    expect(findQueryRow(msa, 'MSGRGKGG').match?.name).toBe('sp_H4copy');
+    expect(findQueryRow(msa, 'MSGRGKGG', 'sp_His4r').match?.name).toBe('sp_His4r');
+});
+test('a named row that matches worse than another does not win', () => {
+    const msa = '>sp_exact\nMSGRGKGG\n>sp_named\nMSGRGKAA\n';
+    expect(findQueryRow(msa, 'MSGRGKGG', 'sp_named').match?.name).toBe('sp_exact');
+});

package/dist/LaunchMsaView/featureMatchesId.test.d.ts

@@ -0,0 +1 @@
+export {};

package/dist/LaunchMsaView/featureMatchesId.test.js

@@ -0,0 +1,74 @@
+import { SimpleFeature } from '@jbrowse/core/util';
+import { expect, test } from 'vitest';
+import { featureMatchesId, findDatasetId } from './util';
+const transcript = new SimpleFeature({
+    uniqueId: 'u1',
+    refName: 'chr17',
+    start: 0,
+    end: 10,
+    id: 'ENST00000269305.9',
+    transcript_id: 'ENST00000269305.9',
+    gene_name: 'TP53',
+});
+test('a transcript matches a dataset keyed by transcript id', () => {
+    expect(featureMatchesId(transcript, 'ENST00000269305.9')).toBe(true);
+});
+test('a transcript matches a dataset keyed by its gene symbol', () => {
+    expect(featureMatchesId(transcript, 'TP53')).toBe(true);
+    expect(featureMatchesId(transcript, 'TP63')).toBe(false);
+});
+// the shape of NCBI's GFF3, as the OrthoFinder synteny tutorial's flies have it
+const ncbiTranscript = new SimpleFeature({
+    uniqueId: 'u2',
+    refName: '2L',
+    start: 9839,
+    end: 18570,
+    id: 'rna-NM_078715.4',
+    name: 'NM_078715.4',
+    subfeatures: [
+        {
+            uniqueId: 'u3',
+            refName: '2L',
+            start: 9900,
+            end: 10000,
+            type: 'CDS',
+            protein_id: 'NP_523439.2',
+        },
+    ],
+});
+const ncbiGene = new SimpleFeature({
+    uniqueId: 'u4',
+    refName: '2L',
+    start: 9839,
+    end: 21376,
+    id: 'gene-Dmel_CG2671',
+    name: 'l(2)gl',
+});
+test("a transcript matches a proteome keyed by its CDS's protein_id", () => {
+    expect(featureMatchesId(ncbiTranscript, 'NP_523439.2')).toBe(true);
+});
+test('a set keyed by gene id matches through the gene, without its gene- prefix', () => {
+    expect(findDatasetId(ncbiTranscript, ncbiGene, new Set(['Dmel_CG2671']))).toBe('Dmel_CG2671');
+    const ensemblGene = new SimpleFeature({
+        uniqueId: 'u5',
+        refName: '17',
+        start: 0,
+        end: 10,
+        id: 'gene:ENSG00000141510',
+    });
+    expect(findDatasetId(transcript, ensemblGene, new Set(['ENSG00000141510']))).toBe('ENSG00000141510');
+});
+test("the transcript's own entry wins over its gene's", () => {
+    expect(findDatasetId(ncbiTranscript, ncbiGene, new Set(['Dmel_CG2671', 'NM_078715.4']))).toBe('NM_078715.4');
+    expect(findDatasetId(ncbiTranscript, ncbiGene, new Set(['other']))).toBe(undefined);
+});
+test('a gene named by symbol matches the name OrthoFinder rewrote', () => {
+    const gene = new SimpleFeature({
+        uniqueId: 'u6',
+        refName: '3R',
+        start: 0,
+        end: 10,
+        id: 'gene-E(spl)m2-BFM',
+    });
+    expect(findDatasetId(ncbiTranscript, gene, new Set(['E_spl_m2-BFM']))).toBe('E_spl_m2-BFM');
+});

package/dist/LaunchMsaView/useQueryRowName.d.ts

@@ -18,7 +18,7 @@ export declare function resolveQueryRowName(queryRow: {
     querySeqName: string;
     names: string[];
 }, fallback: string): string;
-export declare function useQueryRowName(msaText: string, proteinSequence: string): {
+export declare function useQueryRowName(msaText: string, proteinSequence: string, preferredName?: string): {
     detected: import("./detectQueryRow").QueryRowMatch | undefined;
     names: string[];
     querySeqName: string;