@jbrowse/plugin-gtf 2.6.1 → 2.6.3

package/src/GtfAdapter/configSchema.ts

@@ -1,23 +0,0 @@
-import { ConfigurationSchema } from '@jbrowse/core/configuration'
-
-/**
- * #config GtfAdapter
- * #category adapter
- */
-function x() {} // eslint-disable-line @typescript-eslint/no-unused-vars
-
-const GtfAdapter = ConfigurationSchema(
-  'GtfAdapter',
-  {
-    /**
-     * #slot
-     */
-    gtfLocation: {
-      type: 'fileLocation',
-      defaultValue: { uri: '/path/to/my.gtf', locationType: 'UriLocation' },
-    },
-  },
-  { explicitlyTyped: true },
-)
-
-export default GtfAdapter

package/src/GtfAdapter/index.ts

@@ -1 +0,0 @@
-export { default as configSchema } from './configSchema'

package/src/declare.d.ts

@@ -1 +0,0 @@
-declare module '@gmod/gtf'

package/src/index.ts

@@ -1,48 +0,0 @@
-import AdapterType from '@jbrowse/core/pluggableElementTypes/AdapterType'
-import PluginManager from '@jbrowse/core/PluginManager'
-import Plugin from '@jbrowse/core/Plugin'
-import { configSchema as gtfAdapterConfigSchema } from './GtfAdapter'
-import { FileLocation } from '@jbrowse/core/util/types'
-import { AdapterGuesser, getFileName } from '@jbrowse/core/util/tracks'
-
-export default class GtfPlugin extends Plugin {
-  name = 'GTFPlugin'
-
-  install(pluginManager: PluginManager) {
-    pluginManager.addAdapterType(
-      () =>
-        new AdapterType({
-          name: 'GtfAdapter',
-          displayName: 'GTF adapter',
-          configSchema: gtfAdapterConfigSchema,
-          getAdapterClass: () =>
-            import('./GtfAdapter/GtfAdapter').then(r => r.default),
-        }),
-    )
-    pluginManager.addToExtensionPoint(
-      'Core-guessAdapterForLocation',
-      (adapterGuesser: AdapterGuesser) => {
-        return (
-          file: FileLocation,
-          index?: FileLocation,
-          adapterHint?: string,
-        ) => {
-          const regexGuess = /\.gtf(\.gz)?$/i
-          const adapterName = 'GtfAdapter'
-          const fileName = getFileName(file)
-
-          const obj = {
-            type: adapterName,
-            gtfLocation: file,
-          }
-          if (regexGuess.test(fileName) && !adapterHint) {
-            return obj
-          } else if (adapterHint === adapterName) {
-            return obj
-          }
-          return adapterGuesser(file, index, adapterHint)
-        }
-      },
-    )
-  }
-}

package/src/test_data/demo.gtf

@@ -1,26 +0,0 @@
-GeneScaffold_1	ensembl	transcript	107689	107987	.	+	.	gene_id "ENSVPAG00000000407"; gene_version "1"; transcript_id "ENSVPAT00000000407"; transcript_version "1"; gene_source "ensembl"; gene_biotype "pseudogene"; transcript_source "ensembl"; transcript_biotype "pseudogene";
-GeneScaffold_1	ensembl	exon	107689	107715	.	+	.	gene_id "ENSVPAG00000000407"; gene_version "1"; transcript_id "ENSVPAT00000000407"; transcript_version "1"; exon_number "1"; gene_source "ensembl"; gene_biotype "pseudogene"; transcript_source "ensembl"; transcript_biotype "pseudogene"; exon_id "ENSVPAE00000004844"; exon_version "1";
-GeneScaffold_1	ensembl	exon	107717	107812	.	+	.	gene_id "ENSVPAG00000000407"; gene_version "1"; transcript_id "ENSVPAT00000000407"; transcript_version "1"; exon_number "2"; gene_source "ensembl"; gene_biotype "pseudogene"; transcript_source "ensembl"; transcript_biotype "pseudogene"; exon_id "ENSVPAE00000004847"; exon_version "1";
-GeneScaffold_1	ensembl	exon	107824	107877	.	+	.	gene_id "ENSVPAG00000000407"; gene_version "1"; transcript_id "ENSVPAT00000000407"; transcript_version "1"; exon_number "3"; gene_source "ensembl"; gene_biotype "pseudogene"; transcript_source "ensembl"; transcript_biotype "pseudogene"; exon_id "ENSVPAE00000004849"; exon_version "1";
-GeneScaffold_1	ensembl	exon	107881	107895	.	+	.	gene_id "ENSVPAG00000000407"; gene_version "1"; transcript_id "ENSVPAT00000000407"; transcript_version "1"; exon_number "4"; gene_source "ensembl"; gene_biotype "pseudogene"; transcript_source "ensembl"; transcript_biotype "pseudogene"; exon_id "ENSVPAE00000156282"; exon_version "1";
-GeneScaffold_1	ensembl	exon	107900	107917	.	+	.	gene_id "ENSVPAG00000000407"; gene_version "1"; transcript_id "ENSVPAT00000000407"; transcript_version "1"; exon_number "5"; gene_source "ensembl"; gene_biotype "pseudogene"; transcript_source "ensembl"; transcript_biotype "pseudogene"; exon_id "ENSVPAE00000188321"; exon_version "1";
-GeneScaffold_1	ensembl	exon	107919	107987	.	+	.	gene_id "ENSVPAG00000000407"; gene_version "1"; transcript_id "ENSVPAT00000000407"; transcript_version "1"; exon_number "6"; gene_source "ensembl"; gene_biotype "pseudogene"; transcript_source "ensembl"; transcript_biotype "pseudogene"; exon_id "ENSVPAE00000004854"; exon_version "1";
-GeneScaffold_10	ensembl	transcript	37196	42373	.	-	.	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding";
-GeneScaffold_10	ensembl	exon	42303	42373	.	-	.	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "1"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding"; exon_id "ENSVPAE00000120008"; exon_version "1";
-GeneScaffold_10	ensembl	CDS	42303	42373	.	-	0	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "1"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding"; protein_id "ENSVPAP00000009285"; protein_version "1";
-GeneScaffold_10	ensembl	start_codon	42371	42373	.	-	0	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "1"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding";
-GeneScaffold_10	ensembl	exon	41909	42064	.	-	.	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "2"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding"; exon_id "ENSVPAE00000120009"; exon_version "1";
-GeneScaffold_10	ensembl	CDS	41909	42064	.	-	1	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "2"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding"; protein_id "ENSVPAP00000009285"; protein_version "1";
-GeneScaffold_10	ensembl	exon	41149	41278	.	-	.	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "3"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding"; exon_id "ENSVPAE00000120011"; exon_version "1";
-GeneScaffold_10	ensembl	CDS	41149	41278	.	-	1	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "3"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding"; protein_id "ENSVPAP00000009285"; protein_version "1";
-GeneScaffold_10	ensembl	exon	41086	41147	.	-	.	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "4"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding"; exon_id "ENSVPAE00000120015"; exon_version "1";
-GeneScaffold_10	ensembl	CDS	41086	41147	.	-	0	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "4"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding"; protein_id "ENSVPAP00000009285"; protein_version "1";
-GeneScaffold_10	ensembl	exon	40676	40778	.	-	.	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "5"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding"; exon_id "ENSVPAE00000120018"; exon_version "1";
-GeneScaffold_10	ensembl	CDS	40676	40778	.	-	1	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "5"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding"; protein_id "ENSVPAP00000009285"; protein_version "1";
-GeneScaffold_10	ensembl	exon	40215	40368	.	-	.	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "6"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding"; exon_id "ENSVPAE00000120019"; exon_version "1";
-GeneScaffold_10	ensembl	CDS	40215	40368	.	-	0	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "6"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding"; protein_id "ENSVPAP00000009285"; protein_version "1";
-GeneScaffold_10	ensembl	exon	37995	38136	.	-	.	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "7"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding"; exon_id "ENSVPAE00000120020"; exon_version "1";
-GeneScaffold_10	ensembl	CDS	37995	38136	.	-	2	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "7"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding"; protein_id "ENSVPAP00000009285"; protein_version "1";
-GeneScaffold_10	ensembl	exon	37196	37295	.	-	.	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "8"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding"; exon_id "ENSVPAE00000120021"; exon_version "1";
-GeneScaffold_10	ensembl	CDS	37199	37295	.	-	1	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "8"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding"; protein_id "ENSVPAP00000009285"; protein_version "1";
-GeneScaffold_10	ensembl	stop_codon	37196	37198	.	-	0	gene_id "ENSVPAG00000009976"; gene_version "1"; transcript_id "ENSVPAT00000009977"; transcript_version "1"; exon_number "8"; gene_name "RNPS1"; gene_source "ensembl"; gene_biotype "protein_coding"; transcript_name "RNPS1-201"; transcript_source "ensembl"; transcript_biotype "protein_coding";

package/src/test_data/volvox.sorted.gtf

@@ -1,40 +0,0 @@
-ctgA	example	exon	1050	1500	.	+	.	transcript_id "EDEN.1"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	exon	1050	1500	.	+	.	transcript_id "EDEN.2"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	five_prime_UTR	1050	1200	.	+	0	transcript_id "EDEN.1"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	five_prime_UTR	1050	1200	.	+	0	transcript_id "EDEN.2"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	CDS	1201	1500	.	+	0	transcript_id "EDEN.1"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	CDS	1201	1500	.	+	0	transcript_id "EDEN.2"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	exon	1300	1500	.	+	.	transcript_id "EDEN.3"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	five_prime_UTR	1300	1500	.	+	1	transcript_id "EDEN.3"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	CDS	3000	3902	.	+	0	transcript_id "EDEN.1"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	exon	3000	3902	.	+	.	transcript_id "EDEN.1"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	exon	3000	3902	.	+	.	transcript_id "EDEN.3"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	five_prime_UTR	3000	3300	.	+	1	transcript_id "EDEN.3"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	CDS	3301	3902	.	+	0	transcript_id "EDEN.3"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	CDS	5000	5500	.	+	0	transcript_id "EDEN.1"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	CDS	5000	5500	.	+	0	transcript_id "EDEN.2"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	CDS	5000	5500	.	+	1	transcript_id "EDEN.3"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	exon	5000	5500	.	+	.	transcript_id "EDEN.1"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	exon	5000	5500	.	+	.	transcript_id "EDEN.2"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	exon	5000	5500	.	+	.	transcript_id "EDEN.3"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	CDS	7000	7600	.	+	1	transcript_id "EDEN.3"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	CDS	7000	7608	.	+	0	transcript_id "EDEN.1"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	CDS	7000	7608	.	+	0	transcript_id "EDEN.2"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	exon	7000	9000	.	+	.	transcript_id "EDEN.1"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	exon	7000	9000	.	+	.	transcript_id "EDEN.2"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	exon	7000	9000	.	+	.	transcript_id "EDEN.3"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	three_prime_UTR	7601	9000	.	+	1	transcript_id "EDEN.3"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	three_prime_UTR	7609	9000	.	+	0	transcript_id "EDEN.1"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	example	three_prime_UTR	7609	9000	.	+	0	transcript_id "EDEN.2"; gene_id "EDEN"; gene_name "EDEN";
-ctgA	bare_predicted	CDS	10000	11500	.	+	0	transcript_id "Apple1";
-ctgA	predicted	CDS	13500	13800	.	+	0	transcript_id "cds-Apple2";
-ctgA	predicted	CDS	15000	15500	.	+	1	transcript_id "cds-Apple2";
-ctgA	predicted	CDS	16500	17000	.	+	2	transcript_id "cds-Apple2";
-ctgA	exonerate	UTR	17400	17999	.	+	0	transcript_id "rna-Apple3";
-ctgA	exonerate	exon	17400	18800	.	+	.	transcript_id "rna-Apple3";
-ctgA	exonerate	CDS	18000	18800	.	+	0	transcript_id "rna-Apple3";
-ctgA	exonerate	CDS	19000	19500	.	+	0	transcript_id "rna-Apple3";
-ctgA	exonerate	exon	19000	19500	.	+	.	transcript_id "rna-Apple3";
-ctgA	exonerate	CDS	21000	21200	.	+	0	transcript_id "rna-Apple3";
-ctgA	exonerate	exon	21000	23000	.	+	.	transcript_id "rna-Apple3";
-ctgA	exonerate	UTR	21201	23000	.	+	0	transcript_id "rna-Apple3";

package/src/util.ts

@@ -1,77 +0,0 @@
-export type Strand = '+' | '-' | '.' | '?'
-export interface FeatureLoc {
-  [key: string]: unknown
-  start: number
-  end: number
-  strand: Strand
-  seq_name: string
-  child_features: FeatureLoc[][]
-  data: unknown
-  derived_features: unknown
-  attributes: { [key: string]: unknown[] }
-}
-
-export function featureData(data: FeatureLoc) {
-  const f: Record<string, unknown> = { ...data }
-  ;(f.start as number) -= 1 // convert to interbase
-  f.strand = { '+': 1, '-': -1, '.': 0, '?': undefined }[data.strand] // convert strand
-  f.phase = Number(data.frame)
-  f.refName = data.seq_name
-  if (data.score === null) {
-    delete f.score
-  }
-  if (data.frame === null) {
-    delete f.score
-  }
-  const defaultFields = new Set([
-    'start',
-    'end',
-    'seq_name',
-    'score',
-    'featureType',
-    'source',
-    'frame',
-    'strand',
-  ])
-  for (const a of Object.keys(data.attributes)) {
-    let b = a.toLowerCase()
-    if (defaultFields.has(b)) {
-      // add "suffix" to tag name if it already exists
-      // reproduces behavior of NCList
-      b += '2'
-    }
-    if (data.attributes[a] !== null) {
-      let attr = data.attributes[a] as string[] | string
-      if (Array.isArray(attr) && attr.length === 1) {
-        // gtf uses double quotes for text values in the attributes column,
-        // remove them
-        attr = `${attr[0]}`.replaceAll(/^"|"$/g, '')
-      }
-      f[b] = attr
-    }
-  }
-  f.refName = f.seq_name
-  f.type = f.featureType
-
-  // the SimpleFeature constructor takes care of recursively inflating subfeatures
-  if (data.child_features && data.child_features.length > 0) {
-    f.subfeatures = data.child_features.flatMap(childLocs =>
-      childLocs.map(childLoc => featureData(childLoc)),
-    )
-  }
-
-  delete f.child_features
-  delete f.data
-  delete f.derived_features
-  // eslint-disable-next-line no-underscore-dangle
-  delete f._linehash
-  delete f.attributes
-  delete f.seq_name
-  delete f.featureType
-  delete f.frame
-
-  if (f.transcript_id) {
-    f.name = f.transcript_id
-  }
-  return f
-}