jbrowse-plugin-msaview 5.15.0 → 5.16.1

package/README.md

@@ -79,6 +79,8 @@ Linking and embedding:
 - [Launch parameters](docs/launch-parameters.md): every argument and field.
 - [Linked views](docs/linked-views.md): how a column reaches its codon and a
   protein3d structure.
+- [Reading the view](docs/reading-the-view.md): `view.report` and `notices`,
+  what a launch resolved to, as data a script or an agent reads.
 
 Working on the plugin:
 

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';
@@ -33,17 +33,20 @@ const PreLoadedMSA = observer(function ({ model, feature, handleClose, preferred
     const [selectedDatasetId, setSelectedDatasetId] = useState(datasets?.[0]?.datasetId);
     const selectedDataset = datasets?.find(d => d.datasetId === selectedDatasetId);
     const { data: msaList, isLoading: msaListLoading, error: msaListFetchError, } = useFetch(selectedDataset ? `${selectedDataset.datasetId}-msa-list` : null, () => fetchMSAList({ config: selectedDataset.adapter, pluginManager }));
+    const msaIds = useMemo(() => msaList && new Set(msaList), [msaList]);
     const transcriptSelection = useTranscriptSelection({
         feature,
         view,
-        validIds: msaList,
+        hasData: msaIds
+            ? transcript => !!findDatasetId(transcript, feature, msaIds)
+            : undefined,
         preferredTranscriptId,
     });
     const { selectedTranscript, proteinSequence, sequenceStatus } = transcriptSelection;
     // the dataset's name for this transcript, which a track is free not to use
     // as the feature's own id
-    const msaId = selectedTranscript
-        ? msaList?.find(id => featureMatchesId(selectedTranscript, id))
+    const msaId = selectedTranscript && msaIds
+        ? findDatasetId(selectedTranscript, feature, msaIds)
         : undefined;
     const { data: msaData, isLoading: msaDataLoading, error: msaDataFetchError, } = useFetch(msaId && selectedDataset
         ? `${selectedDataset.datasetId}-${msaId}-msa`
@@ -52,8 +55,7 @@ const PreLoadedMSA = observer(function ({ model, feature, handleClose, preferred
         config: selectedDataset.adapter,
         pluginManager,
     }));
-    const msaText = useMemo(() => msaData?.map(r => `>${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
@@ -81,11 +83,11 @@ 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 +102,7 @@ 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

@@ -1,6 +1,9 @@
 import type PluginManager from '@jbrowse/core/PluginManager';
 import type { AnyConfigurationModel } from '@jbrowse/core/configuration';
-import type { Feature } from '@jbrowse/core/util';
+export interface DatasetMsa {
+    msa: string;
+    tree?: string;
+}
 export declare function fetchMSAList({ config, pluginManager, }: {
     config: AnyConfigurationModel;
     pluginManager: PluginManager;
@@ -9,4 +12,4 @@ export declare function fetchMSA({ config, pluginManager, msaId, }: {
     config: AnyConfigurationModel;
     pluginManager: PluginManager;
     msaId: string;
-}): Promise<Feature[]>;
+}): Promise<DatasetMsa>;

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/useTranscriptSelection.d.ts

@@ -1,11 +1,12 @@
 import type { SequenceStatus } from './SequenceStatus';
 import type { Feature } from '@jbrowse/core/util';
-export declare function useTranscriptSelection({ feature, view, validIds, preferredTranscriptId, }: {
+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 +22,5 @@ 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.
+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/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/util.d.ts

@@ -8,6 +8,8 @@ export declare function getTranscriptLength(feature: Feature): {
 };
 export declare function getId(val?: Feature): string;
 export declare function featureMatchesId(feature: Feature, id: string): boolean;
+/** the dataset entry for this transcript, else for the gene it belongs to */
+export declare function findDatasetId(transcript: Feature, gene: Feature, ids: ReadonlySet<string>): string | undefined;
 export declare function getTranscriptDisplayName(val?: Feature): string;
 export declare function getGeneDisplayName(val?: Feature): string;
 export declare function getBlastViewTitle(feature: Feature, transcript: Feature): string;

package/dist/LaunchMsaView/util.js

@@ -39,6 +39,13 @@ export function getTranscriptLength(feature) {
 export function getId(val) {
     return val?.id() ?? '';
 }
+// a proteome downloaded from NCBI or Ensembl names each protein by the CDS's
+// protein_id rather than by its transcript
+function getCdsProteinIds(transcript) {
+    return (transcript.get('subfeatures') ?? [])
+        .filter(f => f.get('type') === 'CDS')
+        .map(f => f.get('protein_id'));
+}
 function getMatchableIds(val) {
     return val
         ? uniqueDefined([
@@ -46,12 +53,43 @@ function getMatchableIds(val) {
             val.get('name'),
             val.get('id'),
             val.get('transcript_id'),
+            val.get('gene_name'),
+            val.get('locus_tag'),
+            ...getCdsProteinIds(val),
         ])
         : [];
 }
 export function featureMatchesId(feature, id) {
     return getMatchableIds(feature).includes(id);
 }
+const geneIdPrefix = /^gene[-:]/;
+// A set keyed by gene, such as an OrthoFinder run over one protein per gene,
+// names the gene by its GFF3 ID without NCBI's `gene-` or Ensembl's `gene:`
+function getGeneMatchableIds(gene) {
+    const id = gene.get('id');
+    return uniqueDefined([
+        id,
+        id?.replace(geneIdPrefix, ''),
+        gene.get('gene_id'),
+        gene.get('locus_tag'),
+    ]);
+}
+// OrthoFinder writes every gene name with each character that Newick or a TSV
+// cannot hold replaced by `_` (its utils/gene_ids.py clean_name), so NCBI's
+// gene `E(spl)m2-BFM` is `E_spl_m2-BFM` in its results. It also replaces control
+// characters, which a GFF3 attribute cannot carry unescaped.
+const orthoFinderUnsafe = /[\s:;(),[\]='"]/g;
+function findIn(candidates, ids) {
+    return (candidates.find(id => ids.has(id)) ??
+        candidates
+            .map(id => id.replace(orthoFinderUnsafe, '_'))
+            .find(id => ids.has(id)));
+}
+/** the dataset entry for this transcript, else for the gene it belongs to */
+export function findDatasetId(transcript, gene, ids) {
+    return (findIn(getMatchableIds(transcript), ids) ??
+        findIn(getGeneMatchableIds(gene), ids));
+}
 export function getTranscriptDisplayName(val) {
     return val ? joinDefined(' ', [val.get('name'), val.get('id')]) : '';
 }

package/dist/MsaViewPanel/afterCreateAutoruns.d.ts

@@ -8,6 +8,18 @@ export declare function loadStoredData(self: JBrowsePluginMsaViewModel): void;
  * refusing IndexedDB fails once rather than in a loop.
  */
 export declare function storeDataToIndexedDB(self: JBrowsePluginMsaViewModel): void;
+interface TranscriptWait {
+    connectedTranscript?: string;
+    connectedFeature?: unknown;
+    blastParams?: unknown;
+    orthologParams?: unknown;
+    launchCompleted: boolean;
+}
+/**
+ * A launch that has not run yet and is held back by the transcript lookup, so
+ * a lookup that fails is a launch that cannot proceed.
+ */
+export declare function launchAwaitsTranscript(self: TranscriptWait): boolean;
 export declare function launchOrthologsIfNeeded(self: JBrowsePluginMsaViewModel): void;
 export declare function launchBlastIfNeeded(self: JBrowsePluginMsaViewModel): void;
 /**
@@ -57,4 +69,10 @@ export declare function syncGenomeHoverToMsaColumn(self: JBrowsePluginMsaViewMod
  */
 export declare function observeProteinHighlights(self: JBrowsePluginMsaViewModel): () => void;
 export declare function runCleanup(): void;
+/**
+ * Waits for the genome view to show an assembly, since the track is added to
+ * that assembly: before navigation `showDomainsOnGenome` has none to name, and
+ * clearing the request then dropped it unfulfilled.
+ */
 export declare function showDomainsOnGenomeIfAsked(self: JBrowsePluginMsaViewModel): void;
+export {};

package/dist/MsaViewPanel/afterCreateAutoruns.js

@@ -11,7 +11,7 @@ import { cleanupOldData, generateDataStoreId, retrieveMsaData, storeMsaData, } f
 import { runLaunch } from './runLaunch';
 import { getProteinViews } from './structureConnection';
 import { proteinPositionsInRange } from './transcriptMap';
-import { getUniprotIdFromAlphaFoldUrl, hasQueryRow, transcriptPosToVisibleCol, } from './util';
+import { errorMessage, getUniprotIdFromAlphaFoldUrl, hasQueryRow, transcriptPosToVisibleCol, } from './util';
 const EXPIRED_MESSAGE = "This view's alignment is no longer in browser storage. Stored alignments are kept for 7 days after they were last used, and are lost when site data is cleared.";
 const RELAUNCHABLE = ' Retry runs the original search again and rebuilds it.';
 const START_OVER = ' Relaunch it from the gene to rebuild it.';
@@ -140,6 +140,15 @@ export function storeDataToIndexedDB(self) {
 function awaitingTranscript(self) {
     return !!self.connectedTranscript && !self.connectedFeature;
 }
+/**
+ * A launch that has not run yet and is held back by the transcript lookup, so
+ * a lookup that fails is a launch that cannot proceed.
+ */
+export function launchAwaitsTranscript(self) {
+    return (!!(self.blastParams ?? self.orthologParams) &&
+        !self.launchCompleted &&
+        awaitingTranscript(self));
+}
 /**
  * The params are the request and `launchCompleted` marks it done. They stay in
  * place on failure, so the error is attributable, and on success, because a
@@ -180,22 +189,30 @@ export function autoLoadProteinDomains(self) {
         };
         addDisposer(self, abort);
         self.setDomainsRequested(true);
+        self.setLoadingDomains(true);
         self.setStatus({
             msg: 'Fetching protein domains from NCBI',
             onCancel: abort,
         });
         void (async () => {
             try {
-                await loadProteinDomains(self, controller.signal);
+                const notice = await loadProteinDomains(self, controller.signal);
+                if (notice && isAlive(self)) {
+                    self.addWarning(notice);
+                }
             }
             catch (e) {
                 if (!isAbortError(e)) {
                     console.error('[msaview-domains] auto-load failed:', e);
+                    if (isAlive(self)) {
+                        self.addWarning(`Protein domains did not load from NCBI: ${errorMessage(e)}`);
+                    }
                 }
             }
             finally {
                 if (isAlive(self)) {
                     self.setStatus(undefined);
+                    self.setLoadingDomains(false);
                 }
             }
         })();
@@ -390,9 +407,14 @@ export function runCleanup() {
         console.error('Failed to cleanup old MSA data:', e);
     });
 }
+/**
+ * Waits for the genome view to show an assembly, since the track is added to
+ * that assembly: before navigation `showDomainsOnGenome` has none to name, and
+ * clearing the request then dropped it unfulfilled.
+ */
 export function showDomainsOnGenomeIfAsked(self) {
     if (self.domainsOnGenome &&
-        self.connectedView &&
+        self.connectedView?.assemblyNames[0] &&
         self.connectedDomains.features.length > 0) {
         self.showDomainsOnGenome();
         self.setDomainsOnGenome(false);

package/dist/MsaViewPanel/displayedRefName.d.ts

@@ -0,0 +1,19 @@
+interface Resolver {
+    initialized?: boolean;
+    getCanonicalRefName: (refName: string) => string | undefined;
+}
+/**
+ * The name a view's displayed regions use for `refName`'s chromosome. core's
+ * bpToPx, and so centerAt, matches region names exactly and does nothing when
+ * the feature says `chr1` and the view displays `1`. The raw name is kept when
+ * a region already carries it, and when no region resolves to the same
+ * chromosome, so a view this cannot place behaves as it did before.
+ */
+export declare function displayedRefName({ displayedRegions, assembly, refName, }: {
+    displayedRegions: readonly {
+        refName: string;
+    }[];
+    assembly?: Resolver;
+    refName: string;
+}): string;
+export {};