jbrowse-plugin-msaview 2.5.2 → 2.6.1

package/dist/version.d.ts

@@ -1 +1 @@
-export declare const version = "2.5.2";
+export declare const version = "2.6.1";

package/dist/version.js

@@ -1 +1 @@
-export const version = '2.5.2';
+export const version = '2.6.1';

package/package.json

@@ -1,5 +1,5 @@
 {
-  "version": "2.5.2",
+  "version": "2.6.1",
   "license": "MIT",
   "name": "jbrowse-plugin-msaview",
   "repository": {
@@ -17,6 +17,7 @@
   ],
   "dependencies": {
     "@emotion/styled": "^11.14.1",
+    "@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

@@ -11,6 +11,8 @@ export default function LaunchMsaViewExtensionPointF(
       session,
       data,
       msaFileLocation,
+      msaIndexedLocation,
+      msaName,
       treeFileLocation,
       connectedViewId,
       connectedFeature,
@@ -29,6 +31,8 @@ export default function LaunchMsaViewExtensionPointF(
       session: AbstractSessionModel
       data?: { msa: string; tree?: string }
       msaFileLocation?: { uri: string }
+      msaIndexedLocation?: { uri: string }
+      msaName?: string
       treeFileLocation?: { uri: string }
       connectedViewId?: string
       connectedFeature?: Record<string, unknown>
@@ -44,7 +48,7 @@ export default function LaunchMsaViewExtensionPointF(
       querySeqName?: string
       highlightColumns?: number[]
     }) => {
-      if (!data && !msaFileLocation) {
+      if (!data && !msaFileLocation && !msaIndexedLocation) {
         throw new Error(
           'No MSA data or file location provided when launching MSA view',
         )
@@ -68,6 +72,8 @@ export default function LaunchMsaViewExtensionPointF(
           msaData: data?.msa,
           treeData: data?.tree,
           msaUrl: msaFileLocation?.uri,
+          msaIndexedLocation,
+          msaName,
           treeUrl: treeFileLocation?.uri,
           querySeqName,
         },

package/src/MsaViewPanel/afterCreateAutoruns.ts

@@ -1,6 +1,7 @@
 import { getSession } from '@jbrowse/core/util'
 
 import { doLaunchBlast } from './doLaunchBlast'
+import { fetchIndexedMsa } from './fetchIndexedMsa'
 import { genomeToMSA } from './genomeToMSA'
 import { loadProteinDomains } from './loadProteinDomains'
 import {
@@ -133,7 +134,15 @@ export function processInit(self: JBrowsePluginMsaViewModel) {
     void (async () => {
       try {
         self.setError(undefined)
-        const { msaData, msaUrl, treeData, treeUrl, querySeqName } = init
+        const {
+          msaData,
+          msaUrl,
+          msaIndexedLocation,
+          msaName,
+          treeData,
+          treeUrl,
+          querySeqName,
+        } = init
 
         if (msaUrl) {
           const id = getUniprotIdFromAlphaFoldUrl(msaUrl)
@@ -156,6 +165,18 @@ export function processInit(self: JBrowsePluginMsaViewModel) {
           }
           const data = await response.text()
           self.setMSA(data)
+        } 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}`,
+            )
+          }
         }
 
         if (treeData) {

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

@@ -1,6 +1,13 @@
 export interface MsaViewInitState {
   msaData?: string
   msaUrl?: 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 = '2.5.2'
+export const version = '2.6.1'