Split VCF

plugins/variants/src/SplitVcfTabixAdapter/SplitVcfTabixAdapter.ts

@@ -0,0 +1,111 @@
+import { TabixIndexedFile } from '@gmod/tabix'
+import VcfParser from '@gmod/vcf'
+import { BaseFeatureDataAdapter } from '@jbrowse/core/data_adapters/BaseAdapter'
+import { fetchAndMaybeUnzipText, updateStatus } from '@jbrowse/core/util'
+import { openLocation } from '@jbrowse/core/util/io'
+import { ObservableCreate } from '@jbrowse/core/util/rxjs'
+
+import VcfFeature from '../VcfFeature'
+
+import type { BaseOptions } from '@jbrowse/core/data_adapters/BaseAdapter'
+import type { Feature } from '@jbrowse/core/util'
+import type { NoAssemblyRegion } from '@jbrowse/core/util/types'
+
+export default class SplitVcfTabixAdapter extends BaseFeatureDataAdapter {
+  private async configurePre(refName: string) {
+    const indexType = this.getConf('indexType')
+    const vcfGzLocation = this.getConf('vcfGzLocationMap')[refName]
+    const indexLocation = this.getConf('indexLocationMap')[refName] || {
+      uri: `${vcfGzLocation.uri}.${indexType.toLowerCase()}`,
+    }
+
+    const filehandle = openLocation(vcfGzLocation, this.pluginManager)
+    const isCSI = indexType === 'CSI'
+    const vcf = new TabixIndexedFile({
+      filehandle,
+      csiFilehandle: isCSI
+        ? openLocation(indexLocation, this.pluginManager)
+        : undefined,
+      tbiFilehandle: !isCSI
+        ? openLocation(indexLocation, this.pluginManager)
+        : undefined,
+      chunkCacheSize: 50 * 2 ** 20,
+    })
+
+    return {
+      vcf,
+      parser: new VcfParser({
+        header: await vcf.getHeader(),
+      }),
+    }
+  }
+
+  async configure(refName: string, opts?: BaseOptions) {
+    const { statusCallback = () => {} } = opts || {}
+    return updateStatus('Downloading index', statusCallback, () =>
+      this.configurePre(refName),
+    )
+  }
+
+  public async getRefNames() {
+    return Object.keys(this.getConf('vcfGzLocationMap'))
+  }
+
+  public getFeatures(query: NoAssemblyRegion, opts: BaseOptions = {}) {
+    return ObservableCreate<Feature>(async observer => {
+      const { refName, start, end } = query
+      const { statusCallback = () => {} } = opts
+      const { vcf, parser } = await this.configure(query.refName, opts)
+
+      await updateStatus('Downloading variants', statusCallback, () =>
+        vcf.getLines(refName, start, end, {
+          lineCallback: (line, fileOffset) => {
+            observer.next(
+              new VcfFeature({
+                variant: parser.parseLine(line),
+                parser,
+                id: `${this.id}-vcf-${fileOffset}`,
+              }),
+            )
+          },
+          ...opts,
+        }),
+      )
+      observer.complete()
+    }, opts.stopToken)
+  }
+
+  async getSources() {
+    const conf = this.getConf('samplesTsvLocation')
+    const r = Object.keys(this.getConf('vcfGzLocationMap'))[0]!
+    if (conf.uri === '' || conf.uri === '/path/to/samples.tsv') {
+      const { parser } = await this.configure(r)
+      return parser.samples.map(name => ({
+        name,
+      }))
+    } else {
+      const txt = await fetchAndMaybeUnzipText(
+        openLocation(conf, this.pluginManager),
+      )
+      const lines = txt.split(/\n|\r\n|\r/)
+      const header = lines[0]!.split('\t')
+      const { parser } = await this.configure(r)
+      const s = new Set(parser.samples)
+      return lines
+        .slice(1)
+        .map(line => {
+          const cols = line.split('\t')
+          return {
+            name: cols[0]!,
+            ...Object.fromEntries(
+              // force col 0 to be called name
+              cols.slice(1).map((c, idx) => [header[idx + 1]!, c] as const),
+            ),
+          }
+        })
+        .filter(f => s.has(f.name))
+    }
+  }
+
+  public freeResources(/* { region } */): void {}
+}

plugins/variants/src/SplitVcfTabixAdapter/configSchema.ts

@@ -0,0 +1,52 @@
+import { ConfigurationSchema } from '@jbrowse/core/configuration'
+
+/**
+ * #config VcfTabixAdapter
+ */
+function x() {} // eslint-disable-line @typescript-eslint/no-unused-vars
+
+const SplitVcfTabixAdapter = ConfigurationSchema(
+  'SplitVcfTabixAdapter',
+  {
+    /**
+     * #slot
+     * object like {chr1:{uri:'url to file'}}
+     */
+    vcfGzLocationMap: {
+      type: 'frozen',
+      defaultValue: {},
+    },
+    /**
+     * #slot
+     * object like {chr1:{uri:'url to index'}}
+     */
+    indexLocationMap: {
+      type: 'frozen',
+      defaultValue: {},
+    },
+
+    /**
+     * #slot
+     */
+    indexType: {
+      type: 'string',
+      defaultValue: 'TBI',
+    },
+
+    /**
+     * #slot
+     */
+    samplesTsvLocation: {
+      type: 'fileLocation',
+      defaultValue: {
+        uri: '/path/to/samples.tsv',
+        description:
+          'tsv with header like name\tpopulation\tetc. where the first column is required, and is the sample names',
+        locationType: 'UriLocation',
+      },
+    },
+  },
+  { explicitlyTyped: true },
+)
+
+export default SplitVcfTabixAdapter

plugins/variants/src/SplitVcfTabixAdapter/index.ts

@@ -0,0 +1,20 @@
+import AdapterType from '@jbrowse/core/pluggableElementTypes/AdapterType'
+
+import configSchema from './configSchema'
+
+import type PluginManager from '@jbrowse/core/PluginManager'
+
+export { default as configSchema } from './configSchema'
+
+export default function VcfTabixAdapterF(pluginManager: PluginManager) {
+  pluginManager.addAdapterType(
+    () =>
+      new AdapterType({
+        name: 'SplitVcfTabixAdapter',
+        displayName: 'VCF tabix adapter (split across multiple files)',
+        configSchema,
+        getAdapterClass: () =>
+          import('./SplitVcfTabixAdapter').then(r => r.default),
+      }),
+  )
+}

plugins/variants/src/index.ts

@@ -6,6 +6,7 @@ import MultiLinearVariantDisplayF from './MultiLinearVariantDisplay'
 import LinearVariantMatrixDisplayF from './MultiLinearVariantMatrixDisplay'
 import LinearVariantMatrixRendererF from './MultiLinearVariantMatrixRenderer'
 import MultiVariantRendererF from './MultiLinearVariantRenderer'
+import SplitVcfTabixAdapterF from './SplitVcfTabixAdapter'
 import StructuralVariantChordRendererF from './StructuralVariantChordRenderer'
 import VariantFeatureWidgetF from './VariantFeatureWidget'
 import { MultiVariantGetGenotypeMatrix } from './VariantRPC/MultiVariantGetGenotypeMatrix'
@@ -24,6 +25,7 @@ export default class VariantsPlugin extends Plugin {
   install(pluginManager: PluginManager) {
     VcfAdapterF(pluginManager)
     VcfTabixAdapterF(pluginManager)
+    SplitVcfTabixAdapterF(pluginManager)
     VariantFeatureWidgetF(pluginManager)
     VariantTrackF(pluginManager)
     ExtensionPointsF(pluginManager)