jbrowse-plugin-msaview 2.6.0 → 2.6.1

package/dist/version.d.ts

@@ -1 +1 @@
-export declare const version = "2.6.0";
+export declare const version = "2.6.1";

package/dist/version.js

@@ -1 +1 @@
-export const version = '2.6.0';
+export const version = '2.6.1';

package/package.json

@@ -1,5 +1,5 @@
 {
-  "version": "2.6.0",
+  "version": "2.6.1",
   "license": "MIT",
   "name": "jbrowse-plugin-msaview",
   "repository": {
@@ -17,7 +17,7 @@
   ],
   "dependencies": {
     "@emotion/styled": "^11.14.1",
-    "@gmod/tabix": "^3.0.1",
+    "@gmod/bgzf-filehandle": "^6.2.0",
     "g2p_mapper": "^2.1.5",
     "idb": "^8.0.3",
     "pako-esm2": "^2.0.2",

package/src/LaunchMsaViewExtensionPoint/index.ts

@@ -11,9 +11,8 @@ export default function LaunchMsaViewExtensionPointF(
       session,
       data,
       msaFileLocation,
-      msaTabixLocation,
-      msaIndexLocation,
-      msaId,
+      msaIndexedLocation,
+      msaName,
       treeFileLocation,
       connectedViewId,
       connectedFeature,
@@ -32,9 +31,8 @@ export default function LaunchMsaViewExtensionPointF(
       session: AbstractSessionModel
       data?: { msa: string; tree?: string }
       msaFileLocation?: { uri: string }
-      msaTabixLocation?: { uri: string }
-      msaIndexLocation?: { uri: string }
-      msaId?: string
+      msaIndexedLocation?: { uri: string }
+      msaName?: string
       treeFileLocation?: { uri: string }
       connectedViewId?: string
       connectedFeature?: Record<string, unknown>
@@ -50,7 +48,7 @@ export default function LaunchMsaViewExtensionPointF(
       querySeqName?: string
       highlightColumns?: number[]
     }) => {
-      if (!data && !msaFileLocation && !msaTabixLocation) {
+      if (!data && !msaFileLocation && !msaIndexedLocation) {
         throw new Error(
           'No MSA data or file location provided when launching MSA view',
         )
@@ -74,9 +72,8 @@ export default function LaunchMsaViewExtensionPointF(
           msaData: data?.msa,
           treeData: data?.tree,
           msaUrl: msaFileLocation?.uri,
-          msaTabixLocation,
-          msaIndexLocation,
-          msaId,
+          msaIndexedLocation,
+          msaName,
           treeUrl: treeFileLocation?.uri,
           querySeqName,
         },

package/src/MsaViewPanel/afterCreateAutoruns.ts

@@ -1,7 +1,7 @@
 import { getSession } from '@jbrowse/core/util'
 
 import { doLaunchBlast } from './doLaunchBlast'
-import { fetchTabixMsa } from './fetchTabixMsa'
+import { fetchIndexedMsa } from './fetchIndexedMsa'
 import { genomeToMSA } from './genomeToMSA'
 import { loadProteinDomains } from './loadProteinDomains'
 import {
@@ -137,9 +137,8 @@ export function processInit(self: JBrowsePluginMsaViewModel) {
         const {
           msaData,
           msaUrl,
-          msaTabixLocation,
-          msaIndexLocation,
-          msaId,
+          msaIndexedLocation,
+          msaName,
           treeData,
           treeUrl,
           querySeqName,
@@ -166,24 +165,17 @@ export function processInit(self: JBrowsePluginMsaViewModel) {
           }
           const data = await response.text()
           self.setMSA(data)
-        } else if (msaTabixLocation) {
-          const feature = self.connectedFeature
-          if (feature) {
-            const fasta = await fetchTabixMsa({
-              location: msaTabixLocation,
-              indexLocation: msaIndexLocation,
-              msaId: msaId ?? String(feature.name),
-              refName: String(feature.refName),
-              start: Number(feature.start),
-              end: Number(feature.end),
-            })
-            if (fasta) {
-              self.setMSA(fasta)
-            } else {
-              throw new Error(
-                `No alignment for ${msaId ?? String(feature.name)} in ${msaTabixLocation.uri}`,
-              )
-            }
+        } else if (msaIndexedLocation && msaName) {
+          const fasta = await fetchIndexedMsa({
+            location: msaIndexedLocation,
+            name: msaName,
+          })
+          if (fasta) {
+            self.setMSA(fasta)
+          } else {
+            throw new Error(
+              `No alignment named ${msaName} in ${msaIndexedLocation.uri}`,
+            )
           }
         }
 

package/src/MsaViewPanel/fetchIndexedMsa.ts

@@ -0,0 +1,47 @@
+import { BgzfFilehandle } from '@gmod/bgzf-filehandle'
+import { openLocation } from '@jbrowse/core/util/io'
+
+// Pull one transcript's whole multiple-alignment out of a single bgzip file
+// keyed by NAME. Given the `.fa.gz` uri, the bgzip index (`.gzi`) and name index
+// (`.idx`, a TSV `name<TAB>offset<TAB>length` of each block's uncompressed byte
+// offset + length) are found by suffix. We fetch the small `.idx` once, look up
+// the name, and random-read just that block — already valid FASTA
+// (`>hg38\nSEQ\n>panTro4\nSEQ\n...`). One genome-scale alignment serves any gene
+// with no per-gene files or coordinates. See react-msaview's gene-explorer.
+export async function fetchIndexedMsa({
+  location,
+  name,
+}: {
+  location: { uri: string }
+  name: string
+}) {
+  const open = (uri: string) =>
+    openLocation({ uri, locationType: 'UriLocation' as const })
+
+  const idxText = await open(`${location.uri}.idx`).readFile('utf8')
+  const entry = lookup(idxText, name)
+  if (entry) {
+    const fh = new BgzfFilehandle({
+      filehandle: open(location.uri),
+      gziFilehandle: open(`${location.uri}.gzi`),
+    })
+    const bytes = await fh.read(entry.length, entry.offset)
+    return new TextDecoder().decode(bytes).trim()
+  }
+  return undefined
+}
+
+// ids are matched versionless, so version drift between the alignment build and
+// the live annotation never breaks the lookup
+const versionless = (s: string) => s.replace(/\.\d+$/, '')
+
+function lookup(idxText: string, name: string) {
+  const want = versionless(name)
+  for (const line of idxText.split('\n')) {
+    const [id, offset, length] = line.split('\t')
+    if (id && versionless(id) === want) {
+      return { offset: Number(offset), length: Number(length) }
+    }
+  }
+  return undefined
+}

package/src/MsaViewPanel/types.ts

@@ -1,14 +1,13 @@
 export interface MsaViewInitState {
   msaData?: string
   msaUrl?: string
-  // a tabix file keyed by genomic locus, where each line packs one transcript's
-  // whole multiple-alignment (`name:SEQ;name:SEQ;...`). The transcript's locus
-  // comes from the view's connectedFeature, and `msaId` (default the feature's
-  // name) selects its line. Lets one genome-scale alignment serve any gene
-  // without per-gene files. See react-msaview's gene-explorer.
-  msaTabixLocation?: { uri: string }
-  msaIndexLocation?: { uri: string }
-  msaId?: string
+  // a single bgzip `.fa.gz` of per-transcript FASTA blocks; its `.gzi` and name
+  // index `.idx` (name<TAB>offset<TAB>length) are found by suffix. `msaName`
+  // selects one transcript's block by name (a random read), so one genome-scale
+  // alignment serves any gene without per-gene files or coordinates. See
+  // react-msaview's gene-explorer.
+  msaIndexedLocation?: { uri: string }
+  msaName?: string
   treeData?: string
   treeUrl?: string
   querySeqName?: string

package/src/version.ts

@@ -1 +1 @@
-export const version = '2.6.0'
+export const version = '2.6.1'

package/dist/MsaViewPanel/fetchTabixMsa.d.ts

@@ -1,12 +0,0 @@
-export declare function fetchTabixMsa({ location, indexLocation, msaId, refName, start, end, }: {
-    location: {
-        uri: string;
-    };
-    indexLocation?: {
-        uri: string;
-    };
-    msaId: string;
-    refName: string;
-    start: number;
-    end: number;
-}): Promise<string | undefined>;

package/dist/MsaViewPanel/fetchTabixMsa.js

@@ -1,33 +0,0 @@
-import { TabixIndexedFile } from '@gmod/tabix';
-import { openLocation } from '@jbrowse/core/util/io';
-// Pull one transcript's whole multiple-alignment out of a locus-keyed tabix
-// file. Each line is `refName<TAB>start<TAB>end<TAB>msaId<TAB>packed`, where
-// `packed` is `name:SEQ;name:SEQ;...` — no newlines, so the alignment survives
-// as a single tabix column. We query the transcript's genomic locus, then pick
-// the line whose msaId matches, and rebuild a FASTA string.
-export async function fetchTabixMsa({ location, indexLocation, msaId, refName, start, end, }) {
-    const uri = (loc) => openLocation({ uri: loc.uri, locationType: 'UriLocation' });
-    const file = new TabixIndexedFile({
-        filehandle: uri(location),
-        csiFilehandle: uri(indexLocation ?? { uri: `${location.uri}.csi` }),
-    });
-    const lines = [];
-    await file.getLines(refName, start, end, {
-        lineCallback: line => {
-            lines.push(line);
-        },
-    });
-    const line = lines.find(l => l.split('\t')[3] === msaId);
-    return line ? unpack(line) : undefined;
-}
-function unpack(line) {
-    const packed = line.split('\t')[4] ?? '';
-    return packed
-        .split(';')
-        .filter(Boolean)
-        .map(pair => {
-        const colon = pair.indexOf(':');
-        return `>${pair.slice(0, colon)}\n${pair.slice(colon + 1)}`;
-    })
-        .join('\n');
-}

package/src/MsaViewPanel/fetchTabixMsa.ts

@@ -1,50 +0,0 @@
-import { TabixIndexedFile } from '@gmod/tabix'
-import { openLocation } from '@jbrowse/core/util/io'
-
-// Pull one transcript's whole multiple-alignment out of a locus-keyed tabix
-// file. Each line is `refName<TAB>start<TAB>end<TAB>msaId<TAB>packed`, where
-// `packed` is `name:SEQ;name:SEQ;...` — no newlines, so the alignment survives
-// as a single tabix column. We query the transcript's genomic locus, then pick
-// the line whose msaId matches, and rebuild a FASTA string.
-export async function fetchTabixMsa({
-  location,
-  indexLocation,
-  msaId,
-  refName,
-  start,
-  end,
-}: {
-  location: { uri: string }
-  indexLocation?: { uri: string }
-  msaId: string
-  refName: string
-  start: number
-  end: number
-}) {
-  const uri = (loc: { uri: string }) =>
-    openLocation({ uri: loc.uri, locationType: 'UriLocation' as const })
-  const file = new TabixIndexedFile({
-    filehandle: uri(location),
-    csiFilehandle: uri(indexLocation ?? { uri: `${location.uri}.csi` }),
-  })
-  const lines: string[] = []
-  await file.getLines(refName, start, end, {
-    lineCallback: line => {
-      lines.push(line)
-    },
-  })
-  const line = lines.find(l => l.split('\t')[3] === msaId)
-  return line ? unpack(line) : undefined
-}
-
-function unpack(line: string) {
-  const packed = line.split('\t')[4] ?? ''
-  return packed
-    .split(';')
-    .filter(Boolean)
-    .map(pair => {
-      const colon = pair.indexOf(':')
-      return `>${pair.slice(0, colon)}\n${pair.slice(colon + 1)}`
-    })
-    .join('\n')
-}