@jbrowse/plugin-gff3 5.0.0-beta.1 → 5.0.0-beta.10

package/esm/Gff3Adapter/Gff3Adapter.js

@@ -22,17 +22,16 @@ export default class Gff3Adapter extends BaseFeatureDataAdapter {
             // that resident set small (8.5x on GENCODE-shaped input), and the
             // render path reads only a handful of attributes anyway — see
             // Gff3Feature.
-            (lines, refName) => {
-                const features = parseLinesLazy(lines);
-                // stamped in place rather than through `{...feature, uniqueId}`:
-                // these are freshly parsed objects nobody else holds, and the spread
-                // copied every attribute of every top-level feature in the file to
-                // add one key
-                for (let i = 0; i < features.length; i++) {
-                    features[i].uniqueId = `${this.id}-${refName}-${i}`;
-                }
-                return features;
-            }, 'Parsing GFF data');
+            //
+            // The id the tree entry carries is minted here, beside the feature
+            // rather than stamped onto it, so the parsed feature is the library's
+            // shape and Gff3Feature serializes it identically for both adapters.
+            (lines, refName) => parseLinesLazy(lines).map((feature, i) => ({
+                start: feature.start,
+                end: feature.end,
+                feature,
+                uniqueId: `${this.id}-${refName}-${i}`,
+            })), 'Parsing GFF data');
             return { header: headerLines.join('\n'), intervalTreeMap };
         },
     });
@@ -52,11 +51,14 @@ export default class Gff3Adapter extends BaseFeatureDataAdapter {
             const { intervalTreeMap } = await this.loadData(opts);
             const tree = intervalTreeMap[refName];
             if (tree) {
-                for (const f of tree(opts.statusCallback).search([start, end])) {
-                    observer.next(new Gff3Feature(f, f.uniqueId));
+                for (const { feature, uniqueId } of tree(opts.statusCallback).search([
+                    start,
+                    end,
+                ])) {
+                    observer.next(new Gff3Feature(feature, uniqueId));
                 }
             }
             observer.complete();
-        }, opts.stopToken);
+        }, opts.signal);
     }
 }

package/esm/Gff3Adapter/configSchema.d.ts

@@ -17,18 +17,16 @@ export declare function normalizeSnapshot(snap: Record<string, unknown>): Record
  * ```
  */
 declare const Gff3Adapter: import("@jbrowse/core/configuration/configurationSchema").ConfigurationSchemaType<{
-    /**
-     * #slot
-     * location of the GFF3 file. May be gzipped; it is decompressed and parsed
-     * in full on first use, so the file has to fit in memory.
-     */
+    type: {
+        type: "string";
+    } & {
+        defaultValue: string;
+    };
     readonly gffLocation: {
-        readonly type: "fileLocation";
-        readonly defaultValue: {
-            readonly uri: "/path/to/my.gff";
-            readonly locationType: "UriLocation";
-        };
+        type: "fileLocation";
+    } & {
+        defaultValue: unknown;
     };
-}, import("@jbrowse/core/configuration/configurationSchema").ConfigurationSchemaOptions<undefined, undefined>>;
+}, import("@jbrowse/core/configuration/configurationSchema").ConfigurationSchemaOptions<undefined, undefined, true, import("@jbrowse/core/configuration").ConfigurationSchemaRequirement<Partial<Record<string, string[]>>, string>, any>>;
 export type Gff3AdapterConfig = Instance<typeof Gff3Adapter>;
 export default Gff3Adapter;

package/esm/Gff3Adapter/configSchema.js

@@ -1,14 +1,6 @@
-import { ConfigurationSchema } from '@jbrowse/core/configuration';
+import { ConfigurationSchema, expandUriShorthand, } from '@jbrowse/core/configuration';
 export function normalizeSnapshot(snap) {
-    return snap.uri
-        ? {
-            ...snap,
-            gffLocation: {
-                uri: snap.uri,
-                baseUri: snap.baseUri,
-            },
-        }
-        : snap;
+    return expandUriShorthand(snap, 'gffLocation');
 }
 /**
  * #config Gff3Adapter

package/esm/Gff3Feature.d.ts

@@ -1,9 +1,5 @@
 import type { Feature, SimpleFeatureSerialized } from '@jbrowse/core/util/simpleFeature';
 import type { LazyGffFeature } from 'gff-nostream';
-/** A parsed GFF3 feature carrying the id the adapter minted for it. */
-export type IdentifiedGffFeature = LazyGffFeature & {
-    uniqueId: string;
-};
 /**
  * A {@link Feature} over a GFF3 feature whose attributes are still raw column-9
  * text, resolving each one only when something asks for it.
@@ -26,7 +22,7 @@ export declare class Gff3Feature implements Feature {
     private uniqueId;
     private parentFeature?;
     private inflated?;
-    constructor(data: IdentifiedGffFeature | LazyGffFeature, uniqueId: string, parentFeature?: Feature | undefined);
+    constructor(data: LazyGffFeature, uniqueId: string, parentFeature?: Feature | undefined);
     /**
      * A switch rather than a lookup table or a `data[name]` fallthrough: the
      * fixed columns are read tens of times per feature by the layout pass, and

package/esm/Gff3TabixAdapter/Gff3TabixAdapter.js

@@ -1,13 +1,12 @@
 import { TabixIndexedFile } from '@gmod/tabix';
 import { BaseFeatureDataAdapter, cachedSetup, } from '@jbrowse/core/data_adapters/BaseAdapter';
 import { downloadStatus } from '@jbrowse/core/util';
-import { sharedBgzfWorkerPool } from '@jbrowse/core/util/bgzfWorkerPool';
 import { decompressedBytesBudget } from '@jbrowse/core/util/cacheBudgets';
 import { openLocation, openTabixIndexFilehandle } from '@jbrowse/core/util/io';
 import { doesIntersect2 } from '@jbrowse/core/util/range';
 import { ObservableCreate } from '@jbrowse/core/util/rxjs';
 import { readTabixLinesRedispatched } from '@jbrowse/core/util/tabix';
-import { parseRecordsLazy } from 'gff-nostream';
+import { hasIdAttribute, parseRecordsLazy } from 'gff-nostream';
 import { Gff3Feature } from '../Gff3Feature.js';
 export default class Gff3TabixAdapter extends BaseFeatureDataAdapter {
     configure = cachedSetup({
@@ -17,14 +16,13 @@ export default class Gff3TabixAdapter extends BaseFeatureDataAdapter {
                 filehandle: openLocation(this.getConf('gffGzLocation'), this.pluginManager),
                 ...openTabixIndexFilehandle(this.getConf(['index', 'location']), this.getConf(['index', 'indexType']), this.pluginManager),
                 chunkCacheBudget: decompressedBytesBudget,
-                bgzfWorkerPool: sharedBgzfWorkerPool(),
             });
+            // the index is a whole-file read, so its byte ticks turn the
+            // "Downloading index" label into a determinate bar
+            await gff.getReferenceSequenceNames({ onProgress });
             return {
                 gff,
                 dontRedispatchSet: new Set(this.getConf('dontRedispatch')),
-                // the index is a whole-file read, so its byte ticks turn the
-                // "Downloading index" label into a determinate bar
-                header: await gff.getHeader({ onProgress }),
             };
         },
     });
@@ -33,8 +31,8 @@ export default class Gff3TabixAdapter extends BaseFeatureDataAdapter {
         return downloadStatus('Downloading index', opts.statusCallback, onProgress => gff.getReferenceSequenceNames({ ...opts, onProgress }));
     }
     async getHeader(opts = {}) {
-        const { header } = await this.configure(opts);
-        return header;
+        const { gff } = await this.configure(opts);
+        return gff.getHeader(opts);
     }
     // Index-only compressed-byte estimate (no feature download), used by the
     // feature-fetch RPC to short-circuit an over-budget region before pulling
@@ -47,7 +45,26 @@ export default class Gff3TabixAdapter extends BaseFeatureDataAdapter {
         return ObservableCreate(async (observer) => {
             try {
                 const { gff, dontRedispatchSet } = await this.configure(opts);
-                const lines = await readTabixLinesRedispatched(gff, query, dontRedispatchSet, opts);
+                // The bound exists to reach a record's children, so what disqualifies a
+                // record is having none. `ID` answers that exactly — nothing can name a
+                // parent that has no name — and gff-nostream decides it with the scan
+                // its linker uses, so a record the tree would attach children to is
+                // never one the bound leaves out. `dontRedispatch` can only approximate
+                // the same question by type: every NCBI `GCF_*_genomic.gff.gz` opens
+                // each reference with a chromosome-long `match` record no list named,
+                // and none of them has an `ID`. The list covers what `ID` cannot see: a
+                // record carrying an `ID` that nothing references, which hosted hg19
+                // RefSeq's chromosome-long `region` is.
+                const canHaveChildren = (line) => hasIdAttribute(line.line) && !dontRedispatchSet.has(line.type);
+                const lines = await readTabixLinesRedispatched(gff, query, 
+                // A caller reading only top-level features is owed nothing by the
+                // flanks and pays for them anyway: they complete SUBFEATURE lists, and
+                // a top-level feature overlapping the query is in the query's own read
+                // already, since tabix returns every overlapping line and a child is
+                // contained in its parent. On an NCBI `GCF_*_genomic.gff.gz`, whose
+                // chromosome-long `match` widens the bound at every window, that is
+                // 193,008 lines parsed to keep 3 features — 2734 ms against 8 ms.
+                opts.topLevelOnly ? () => false : canHaveChildren, opts);
                 // emit only top-level features intersecting the original query. the
                 // byte offset stays on our own record and is used purely to mint a
                 // stable id, so it never pollutes the feature's data
@@ -61,6 +78,6 @@ export default class Gff3TabixAdapter extends BaseFeatureDataAdapter {
             catch (e) {
                 observer.error(e);
             }
-        }, opts.stopToken);
+        }, opts.signal);
     }
 }

package/esm/Gff3TabixAdapter/configSchema.d.ts

@@ -18,51 +18,38 @@ export declare function normalizeSnapshot(snap: Record<string, unknown>): Record
  * ```
  */
 declare const Gff3TabixAdapter: import("@jbrowse/core/configuration/configurationSchema").ConfigurationSchemaType<{
-    /**
-     * #slot
-     * location of the bgzip-compressed GFF3, sorted by position. Must be bgzip
-     * rather than plain gzip, which tabix cannot index.
-     */
+    readonly densityAdapter: {
+        type: "maybeFrozen";
+    };
+    type: {
+        type: "string";
+    } & {
+        defaultValue: string;
+    };
     readonly gffGzLocation: {
-        readonly type: "fileLocation";
-        readonly defaultValue: {
-            readonly uri: "/path/to/my.gff.gz";
-            readonly locationType: "UriLocation";
-        };
+        type: "fileLocation";
+    } & {
+        defaultValue: unknown;
     };
     readonly index: import("@jbrowse/core/configuration/configurationSchema").ConfigurationSchemaType<{
         readonly indexType: {
-            readonly model: import("@jbrowse/mobx-state-tree").ISimpleType<"CSI" | "TBI">;
-            readonly type: 'stringEnum';
-            readonly defaultValue: 'TBI';
-            readonly description: '`TBI` is the usual `tabix` output. `CSI` is required for a reference longer than 512 Mb, which TBI cannot address.';
+            type: "stringEnum";
+        } & {
+            model: import("@jbrowse/mobx-state-tree").ISimpleType<"CSI" | "TBI">;
+        } & {
+            defaultValue: string;
         };
         readonly location: {
-            readonly type: 'fileLocation';
-            readonly defaultValue: {
-                readonly uri: '/path/to/my.gz.tbi';
-                readonly locationType: 'UriLocation';
-            };
-            readonly description: 'location of the tabix index. Only needed when it is not named `<file>.tbi` (or `.csi`), which is what the `uri` shorthand assumes.';
+            type: "fileLocation";
+        } & {
+            defaultValue: unknown;
         };
-    }, import("@jbrowse/core/configuration/configurationSchema").ConfigurationSchemaOptions<undefined, undefined>>;
-    /**
-     * #slot
-     * the Gff3TabixAdapter has to "redispatch" if it fetches a region and
-     * features it finds inside that region extend outside the region we requested.
-     * you can disable this for certain feature types to avoid fetching e.g. the
-     * entire chromosome
-     *
-     * the defaults are the whole-sequence records the common GFF3 sources emit:
-     * `region` (NCBI), `supercontig`/`scaffold` (Ensembl, for non-chromosomal
-     * sequences), plus `chromosome` and `contig`. They span an entire reference
-     * and have no children, so letting one expand the fetch would pull a whole
-     * chromosome to gain nothing
-     */
+    }, import("@jbrowse/core/configuration/configurationSchema").ConfigurationSchemaOptions<undefined, undefined, undefined, import("@jbrowse/core/configuration").ConfigurationSchemaRequirement<Partial<Record<string, string[]>>, string>, any>>;
     readonly dontRedispatch: {
-        readonly type: "stringArray";
-        readonly defaultValue: readonly ["chromosome", "region", "contig", "supercontig", "scaffold"];
+        type: "stringArray";
+    } & {
+        defaultValue: unknown;
     };
-}, import("@jbrowse/core/configuration/configurationSchema").ConfigurationSchemaOptions<undefined, undefined>>;
+}, import("@jbrowse/core/configuration/configurationSchema").ConfigurationSchemaOptions<undefined, undefined, true, import("@jbrowse/core/configuration").ConfigurationSchemaRequirement<Partial<Record<string, string[]>>, string>, any>>;
 export type Gff3TabixAdapterConfig = Instance<typeof Gff3TabixAdapter>;
 export default Gff3TabixAdapter;

package/esm/Gff3TabixAdapter/configSchema.js

@@ -1,4 +1,5 @@
-import { ConfigurationSchema, expandTabixShorthand, tabixIndexFields, } from '@jbrowse/core/configuration';
+import { ConfigurationSchema, expandTabixShorthand, tabixIndexSchema, } from '@jbrowse/core/configuration';
+import { densityAdapterConfigSchemaFields } from '@jbrowse/core/data_adapters/BaseAdapter';
 export function normalizeSnapshot(snap) {
     return expandTabixShorthand(snap, 'gffGzLocation');
 }
@@ -32,7 +33,12 @@ const Gff3TabixAdapter = ConfigurationSchema('Gff3TabixAdapter', {
             locationType: 'UriLocation',
         },
     },
-    index: ConfigurationSchema('TabixIndex', { ...tabixIndexFields }),
+    /**
+     * #slot
+     * where the tabix index is and which kind it is. The `uri` shorthand derives
+     * both, so a config using it states neither.
+     */
+    index: tabixIndexSchema(),
     /**
      * #slot
      * the Gff3TabixAdapter has to "redispatch" if it fetches a region and
@@ -56,6 +62,7 @@ const Gff3TabixAdapter = ConfigurationSchema('Gff3TabixAdapter', {
             'scaffold',
         ],
     },
+    ...densityAdapterConfigSchemaFields,
 }, {
     explicitlyTyped: true,
     /**

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jbrowse/plugin-gff3",
-  "version": "5.0.0-beta.1",
+  "version": "5.0.0-beta.10",
   "type": "module",
   "description": "JBrowse 2 gff3.",
   "keywords": [
@@ -21,12 +21,12 @@
     "esm"
   ],
   "dependencies": {
-    "@gmod/tabix": "^3.8.2",
-    "@jbrowse/mobx-state-tree": "^6.5.1",
-    "gff-nostream": "^5.3.0",
-    "mobx": "^7.0.3",
+    "@gmod/tabix": "^3.8.3",
+    "@jbrowse/mobx-state-tree": "^6.6.1",
+    "gff-nostream": "^5.5.0",
+    "mobx": "^7.0.5",
     "rxjs": "^7.8.2",
-    "@jbrowse/core": "^5.0.0-beta.1"
+    "@jbrowse/core": "^5.0.0-beta.10"
   },
   "publishConfig": {
     "access": "public"