Add better support for bedMethyl files (#4662)

packages/core/data_adapters/BaseAdapter/BaseFeatureDataAdapter.ts

@@ -253,4 +253,19 @@ export abstract class BaseFeatureDataAdapter extends BaseAdapter {
     }
     return this.getRegionFeatureDensityStats(regions[0]!, opts)
   }
+
+  async getSources(
+    regions: Region[],
+  ): Promise<{ name: string; color?: string; [key: string]: unknown }[]> {
+    const features = await firstValueFrom(
+      this.getFeaturesInMultipleRegions(regions).pipe(toArray()),
+    )
+    const sources = new Set<string>()
+    for (const f of features) {
+      sources.add(f.get('source'))
+    }
+    return [...sources].map(source => ({
+      name: source,
+    }))
+  }
 }

plugins/bed/src/BedAdapter/BedAdapter.ts

@@ -5,7 +5,12 @@ import {
 } from '@jbrowse/core/data_adapters/BaseAdapter'
 import { openLocation } from '@jbrowse/core/util/io'
 import { ObservableCreate } from '@jbrowse/core/util/rxjs'
-import { Region, Feature, fetchAndMaybeUnzip } from '@jbrowse/core/util'
+import {
+  Region,
+  Feature,
+  fetchAndMaybeUnzip,
+  SimpleFeature,
+} from '@jbrowse/core/util'
 import IntervalTree from '@flatten-js/interval-tree'
 
 // locals
@@ -123,17 +128,19 @@ export default class BedAdapter extends BaseFeatureDataAdapter {
     const names = await this.getNames()
 
     const intervalTree = new IntervalTree()
-    const ret = lines.map((f, i) => {
+    const ret = lines.map((line, i) => {
       const uniqueId = `${this.id}-${refName}-${i}`
-      return featureData(
-        f,
-        colRef,
-        colStart,
-        colEnd,
-        scoreColumn,
-        parser,
-        uniqueId,
-        names,
+      return new SimpleFeature(
+        featureData({
+          line,
+          colRef,
+          colStart,
+          colEnd,
+          scoreColumn,
+          parser,
+          uniqueId,
+          names,
+        }),
       )
     })
 

plugins/bed/src/BedAdapter/__snapshots__/BedAdapter.test.ts.snap

@@ -129,8 +129,15 @@ exports[`adapter can fetch features bed with autosql 1`] = `
     "Variant_Type": "SNP",
     "alcohol_history": "--",
     "alcohol_intensity": "--",
+    "blockCount": 1,
+    "blockSizes": [
+      1,
+    ],
     "bmi": "--",
     "case_id": "09454ed6-64bc-4a35-af44-7c4344623d45",
+    "chromStarts": [
+      0,
+    ],
     "cigarettes_per_day": "--",
     "days_to_death": "--",
     "dbSNP_RS": "novel",
@@ -160,6 +167,8 @@ exports[`adapter can fetch features bed with autosql 1`] = `
         "uniqueId": "test-ctgA-0-0",
       },
     ],
+    "thickEnd": 1815757,
+    "thickStart": 1815756,
     "type": undefined,
     "uniqueId": "test-ctgA-0",
     "weight": "--",

plugins/bed/src/BedTabixAdapter/BedTabixAdapter.ts

@@ -5,7 +5,12 @@ import {
 } from '@jbrowse/core/data_adapters/BaseAdapter'
 import { openLocation } from '@jbrowse/core/util/io'
 import { ObservableCreate } from '@jbrowse/core/util/rxjs'
-import { FileLocation, Region, Feature } from '@jbrowse/core/util'
+import {
+  FileLocation,
+  Region,
+  Feature,
+  SimpleFeature,
+} from '@jbrowse/core/util'
 import { TabixIndexedFile } from '@gmod/tabix'
 import PluginManager from '@jbrowse/core/PluginManager'
 import { AnyConfigurationModel } from '@jbrowse/core/configuration'
@@ -78,22 +83,21 @@ export default class BedTabixAdapter extends BaseFeatureDataAdapter {
       const colRef = columnNumbers.ref - 1
       const colStart = columnNumbers.start - 1
       const colEnd = columnNumbers.end - 1
-      // colSame handles special case for tabix where a single column is both
-      // the start and end, this is assumed to be covering the base at this
-      // position (e.g. tabix -s 1 -b 2 -e 2) begin and end are same
       const names = await this.getNames()
       await this.bed.getLines(query.refName, query.start, query.end, {
         lineCallback: (line, fileOffset) => {
           observer.next(
-            featureData(
-              line,
-              colRef,
-              colStart,
-              colEnd,
-              this.scoreColumn,
-              this.parser,
-              `${this.id}-${fileOffset}`,
-              names,
+            new SimpleFeature(
+              featureData({
+                line,
+                colRef,
+                colStart,
+                colEnd,
+                scoreColumn: this.scoreColumn,
+                parser: this.parser,
+                uniqueId: `${this.id}-${fileOffset}`,
+                names,
+              }),
             ),
           )
         },

plugins/bed/src/BedTabixAdapter/__snapshots__/BedTabixAdapter.test.ts.snap

@@ -129,8 +129,15 @@ exports[`adapter can fetch features bed with autosql 1`] = `
     "Variant_Type": "SNP",
     "alcohol_history": "--",
     "alcohol_intensity": "--",
+    "blockCount": 1,
+    "blockSizes": [
+      1,
+    ],
     "bmi": "--",
     "case_id": "09454ed6-64bc-4a35-af44-7c4344623d45",
+    "chromStarts": [
+      0,
+    ],
     "cigarettes_per_day": "--",
     "days_to_death": "--",
     "dbSNP_RS": "novel",
@@ -160,6 +167,8 @@ exports[`adapter can fetch features bed with autosql 1`] = `
         "uniqueId": "test-52986-0",
       },
     ],
+    "thickEnd": 1815757,
+    "thickStart": 1815756,
     "type": undefined,
     "uniqueId": "test-52986",
     "weight": "--",

plugins/bed/src/BigBedAdapter/BigBedAdapter.ts

@@ -13,18 +13,12 @@ import {
   min,
   Feature,
   SimpleFeature,
-  SimpleFeatureSerializedNoId,
+  SimpleFeatureSerialized,
 } from '@jbrowse/core/util'
 import { firstValueFrom, Observer, toArray } from 'rxjs'
 
 // locals
-import {
-  isUcscProcessedTranscript,
-  ucscProcessedTranscript,
-  makeRepeatTrackDescription,
-  makeBlocks,
-  arrayify,
-} from '../util'
+import { featureData2 } from '../util'
 
 export default class BigBedAdapter extends BaseFeatureDataAdapter {
   private cachedP?: Promise<{
@@ -148,19 +142,14 @@ export default class BigBedAdapter extends BaseFeatureDataAdapter {
       }
     }
 
-    const parentAggregation = {} as Record<
-      string,
-      SimpleFeatureSerializedNoId[]
-    >
+    const parentAggregation = {} as Record<string, SimpleFeatureSerialized[]>
 
     if (feats.some(f => f.uniqueId === undefined)) {
       throw new Error('found uniqueId undefined')
     }
     for (const feat of feats) {
-      const data = parser.parseLine(
-        `${query.refName}\t${feat.start}\t${feat.end}\t${feat.rest}`,
-        { uniqueId: feat.uniqueId! },
-      )
+      const line = `${query.refName}\t${feat.start}\t${feat.end}\t${feat.rest}`
+      const data = parser.parseLine(line, { uniqueId: feat.uniqueId! })
 
       const aggr = data[aggregateField]
       if (!parentAggregation[aggr]) {
@@ -183,89 +172,27 @@ export default class BigBedAdapter extends BaseFeatureDataAdapter {
         strand,
         ...rest
       } = data
-      const chromStarts = arrayify(chromStarts2)
-      const blockStarts = arrayify(blockStarts2)
-      const blockSizes = arrayify(blockSizes2)
-      const score = scoreColumn ? +data[scoreColumn] : +score2
 
-      const subfeatures = makeBlocks({
-        chromStarts,
-        blockStarts,
-        blockSizes,
-        blockCount,
+      const f = featureData2({
+        ...rest,
+        scoreColumn,
+        line,
+        parser,
         uniqueId,
-        refName: query.refName,
         start: feat.start,
+        end: feat.end,
+        refName: query.refName,
       })
-
-      if (
-        isUcscProcessedTranscript({
-          strand,
-          blockCount,
-          thickStart,
-          description,
-        })
-      ) {
-        const f = ucscProcessedTranscript({
-          ...rest,
-          strand,
-          uniqueId,
-          type,
-          start: feat.start,
-          end: feat.end,
-          refName: query.refName,
-          score,
-          description,
-          chromStarts: chromStarts!,
-          blockSizes: blockSizes!,
-          blockCount,
-          thickStart,
-          thickEnd,
-          subfeatures,
-        })
-        if (aggr) {
-          parentAggregation[aggr].push(f)
-        } else {
-          if (
-            doesIntersect2(
-              f.start,
-              f.end,
-              originalQuery.start,
-              originalQuery.end,
-            )
-          ) {
-            observer.next(
-              new SimpleFeature({
-                id: `${this.id}-${uniqueId}`,
-                data: f,
-              }),
-            )
-          }
-        }
+      if (aggr) {
+        parentAggregation[aggr].push(f)
       } else {
         if (
-          doesIntersect2(
-            feat.start,
-            feat.end,
-            originalQuery.start,
-            originalQuery.end,
-          )
+          doesIntersect2(f.start, f.end, originalQuery.start, originalQuery.end)
         ) {
           observer.next(
             new SimpleFeature({
               id: `${this.id}-${uniqueId}`,
-              data: {
-                ...rest,
-                ...makeRepeatTrackDescription(description),
-                start: feat.start,
-                end: feat.end,
-                strand,
-                uniqueId,
-                type,
-                score,
-                refName: query.refName,
-                subfeatures,
-              },
+              data: f,
             }),
           )
         }

plugins/bed/src/generateBedMethylFeature.ts

@@ -0,0 +1,67 @@
+export function isBedMethylFeature({
+  splitLine,
+  start,
+  end,
+}: {
+  splitLine: string[]
+  start: number
+  end: number
+}) {
+  return +(splitLine[6] || 0) === start && +(splitLine[7] || 0) === end
+}
+export function generateBedMethylFeature({
+  line,
+  uniqueId,
+  refName,
+  start,
+  end,
+}: {
+  line: string
+  uniqueId: string
+  refName: string
+  start: number
+  end: number
+}) {
+  // see
+  // https://github.com/nanoporetech/modkit?tab=readme-ov-file#description-of-bedmethyl-output
+  const [
+    ,
+    ,
+    ,
+    code,
+    ,
+    strand,
+    ,
+    ,
+    color,
+    n_valid_cov,
+    fraction_modified,
+    n_mod,
+    n_canonical,
+    n_other_mod,
+    n_delete,
+    n_fail,
+    n_diff,
+    n_nocall,
+  ] = line.split('\t')
+  return {
+    uniqueId,
+    refName,
+    start,
+    end,
+    code,
+    score: fraction_modified,
+    strand,
+    color,
+    source: code,
+    n_valid_cov,
+    fraction_modified,
+    n_mod,
+    n_canonical,
+    n_other_mod,
+    n_delete,
+    n_fail,
+    n_diff,
+    n_nocall,
+  }
+}