@jbrowse/plugin-bed 2.16.0 → 2.17.0

package/esm/BigBedAdapter/BigBedAdapter.js

@@ -4,8 +4,9 @@ import { BaseFeatureDataAdapter, } from '@jbrowse/core/data_adapters/BaseAdapter
 import { openLocation } from '@jbrowse/core/util/io';
 import { ObservableCreate } from '@jbrowse/core/util/rxjs';
 import { doesIntersect2, max, min, SimpleFeature, } from '@jbrowse/core/util';
+import { firstValueFrom, toArray } from 'rxjs';
 // locals
-import { isUcscProcessedTranscript, makeBlocks, ucscProcessedTranscript, } from '../util';
+import { featureData2 } from '../util';
 export default class BigBedAdapter extends BaseFeatureDataAdapter {
     async configurePre(opts) {
         const pm = this.pluginManager;
@@ -13,22 +14,42 @@ export default class BigBedAdapter extends BaseFeatureDataAdapter {
             filehandle: openLocation(this.getConf('bigBedLocation'), pm),
         });
         const header = await bigbed.getHeader(opts);
-        const parser = new BED({ autoSql: header.autoSql });
-        return { bigbed, header, parser };
+        const parser = new BED({
+            autoSql: header.autoSql,
+        });
+        return {
+            bigbed,
+            header,
+            parser,
+        };
     }
     async configure(opts) {
-        if (!this.cached) {
-            this.cached = this.configurePre(opts).catch((e) => {
-                this.cached = undefined;
+        if (!this.cachedP) {
+            this.cachedP = this.configurePre(opts).catch((e) => {
+                this.cachedP = undefined;
                 throw e;
             });
         }
-        return this.cached;
+        return this.cachedP;
     }
     async getRefNames(opts) {
         const { header } = await this.configure(opts);
         return Object.keys(header.refsByName);
     }
+    async getData() {
+        const refNames = await this.getRefNames();
+        const features = [];
+        for (const refName of refNames) {
+            const f = await firstValueFrom(this.getFeatures({
+                assemblyName: 'unknown',
+                refName,
+                start: 0,
+                end: Number.MAX_SAFE_INTEGER,
+            }).pipe(toArray()));
+            features.push(f);
+        }
+        return features.flat();
+    }
     async getHeader(opts) {
         const { parser, header } = await this.configure(opts);
         const { version, fileType } = header;
@@ -40,7 +61,7 @@ export default class BigBedAdapter extends BaseFeatureDataAdapter {
             fields: Object.fromEntries(fields.map(({ name, comment }) => [name, comment])),
         };
     }
-    async getFeaturesHelper(query, opts, observer, allowRedispatch, originalQuery = query) {
+    async getFeaturesHelper({ query, opts, observer, allowRedispatch, originalQuery = query, }) {
         const { signal } = opts;
         const scoreColumn = this.getConf('scoreColumn');
         const aggregateField = this.getConf('aggregateField');
@@ -61,7 +82,17 @@ export default class BigBedAdapter extends BaseFeatureDataAdapter {
                 }
             }
             if (maxEnd > query.end || minStart < query.start) {
-                await this.getFeaturesHelper({ ...query, start: minStart, end: maxEnd }, opts, observer, false, query);
+                await this.getFeaturesHelper({
+                    query: {
+                        ...query,
+                        start: minStart,
+                        end: maxEnd,
+                    },
+                    opts,
+                    observer,
+                    allowRedispatch: false,
+                    originalQuery: query,
+                });
                 return;
             }
         }
@@ -70,62 +101,31 @@ export default class BigBedAdapter extends BaseFeatureDataAdapter {
             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]) {
                 parentAggregation[aggr] = [];
             }
-            const { uniqueId, type, chromStart, chromStarts, blockStarts, blockCount, blockSizes, chromEnd, thickStart, thickEnd, chrom, score, ...rest } = data;
-            const subfeatures = blockCount
-                ? makeBlocks({
-                    chromStarts,
-                    blockStarts,
-                    blockCount,
-                    blockSizes,
-                    uniqueId,
-                    refName: query.refName,
-                    start: feat.start,
-                })
-                : [];
-            if (isUcscProcessedTranscript(data)) {
-                const f = ucscProcessedTranscript({
-                    ...rest,
-                    uniqueId,
-                    type,
-                    start: feat.start,
-                    end: feat.end,
-                    refName: query.refName,
-                    score: scoreColumn ? +data[scoreColumn] : score,
-                    chromStarts,
-                    blockCount,
-                    blockSizes,
-                    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 }));
-                    }
-                }
+            const { uniqueId, type, chrom, chromStart, chromEnd, description, chromStarts: chromStarts2, blockStarts: blockStarts2, blockSizes: blockSizes2, score: score2, blockCount, thickStart, thickEnd, strand, ...rest } = data;
+            const f = featureData2({
+                ...rest,
+                scoreColumn,
+                line,
+                parser,
+                uniqueId,
+                start: feat.start,
+                end: feat.end,
+                refName: query.refName,
+            });
+            if (aggr) {
+                parentAggregation[aggr].push(f);
             }
             else {
-                if (doesIntersect2(feat.start, feat.end, originalQuery.start, originalQuery.end)) {
+                if (doesIntersect2(f.start, f.end, originalQuery.start, originalQuery.end)) {
                     observer.next(new SimpleFeature({
                         id: `${this.id}-${uniqueId}`,
-                        data: {
-                            ...rest,
-                            uniqueId,
-                            type,
-                            start: feat.start,
-                            score: scoreColumn ? +data[scoreColumn] : score,
-                            end: feat.end,
-                            refName: query.refName,
-                            subfeatures,
-                        },
+                        data: f,
                     }));
                 }
             }
@@ -154,7 +154,12 @@ export default class BigBedAdapter extends BaseFeatureDataAdapter {
     getFeatures(query, opts = {}) {
         return ObservableCreate(async (observer) => {
             try {
-                await this.getFeaturesHelper(query, opts, observer, true);
+                await this.getFeaturesHelper({
+                    query,
+                    opts,
+                    observer,
+                    allowRedispatch: true,
+                });
             }
             catch (e) {
                 observer.error(e);

package/esm/generateBedMethylFeature.d.ts

@@ -0,0 +1,31 @@
+export declare function isBedMethylFeature({ splitLine, start, end, }: {
+    splitLine: string[];
+    start: number;
+    end: number;
+}): boolean;
+export declare function generateBedMethylFeature({ line, uniqueId, refName, start, end, }: {
+    line: string;
+    uniqueId: string;
+    refName: string;
+    start: number;
+    end: number;
+}): {
+    uniqueId: string;
+    refName: string;
+    start: number;
+    end: number;
+    code: string | undefined;
+    score: string | undefined;
+    strand: string | undefined;
+    color: string | undefined;
+    source: string | undefined;
+    n_valid_cov: string | undefined;
+    fraction_modified: string | undefined;
+    n_mod: string | undefined;
+    n_canonical: string | undefined;
+    n_other_mod: string | undefined;
+    n_delete: string | undefined;
+    n_fail: string | undefined;
+    n_diff: string | undefined;
+    n_nocall: string | undefined;
+};

package/esm/generateBedMethylFeature.js

@@ -0,0 +1,28 @@
+export function isBedMethylFeature({ splitLine, start, end, }) {
+    return +(splitLine[6] || 0) === start && +(splitLine[7] || 0) === end;
+}
+export function generateBedMethylFeature({ line, uniqueId, refName, start, end, }) {
+    // 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,
+    };
+}

package/esm/generateRepeatMaskerFeature.d.ts

@@ -0,0 +1,51 @@
+export declare function isRepeatMaskerDescriptionField(desc?: string): desc is string;
+export declare function generateRepeatMaskerFeature({ uniqueId, refName, start, end, description, ...rest }: {
+    uniqueId: string;
+    refName: string;
+    start: number;
+    end: number;
+    description: string;
+    [key: string]: unknown;
+}): {
+    uniqueId: string;
+    refName: string;
+    start: number;
+    end: number;
+    bitsw_score: string | undefined;
+    percent_div: string | undefined;
+    percent_del: string | undefined;
+    percent_ins: string | undefined;
+    query_chr: string | undefined;
+    query_begin: string | undefined;
+    query_end: string | undefined;
+    query_remaining: string | undefined;
+    orientation: string | undefined;
+    matching_repeat_name: string | undefined;
+    matching_repeat_class: string | undefined;
+    matching_repeat_begin: string | undefined;
+    matching_repeat_end: string | undefined;
+    matching_repeat_remaining: string | undefined;
+    repeat_id: string | undefined;
+    description?: undefined;
+} | {
+    uniqueId: string;
+    refName: string;
+    start: number;
+    end: number;
+    description: undefined;
+    bitsw_score?: undefined;
+    percent_div?: undefined;
+    percent_del?: undefined;
+    percent_ins?: undefined;
+    query_chr?: undefined;
+    query_begin?: undefined;
+    query_end?: undefined;
+    query_remaining?: undefined;
+    orientation?: undefined;
+    matching_repeat_name?: undefined;
+    matching_repeat_class?: undefined;
+    matching_repeat_begin?: undefined;
+    matching_repeat_end?: undefined;
+    matching_repeat_remaining?: undefined;
+    repeat_id?: undefined;
+};

package/esm/generateRepeatMaskerFeature.js

@@ -0,0 +1,37 @@
+export function isRepeatMaskerDescriptionField(desc) {
+    const ret = desc === null || desc === void 0 ? void 0 : desc.trim().split(' ');
+    return [0, 1, 2, 3, 5, 6].every(s => (ret === null || ret === void 0 ? void 0 : ret[s]) !== undefined ? !Number.isNaN(+ret[s]) : false);
+}
+function makeRepeatTrackDescription(description) {
+    if (isRepeatMaskerDescriptionField(description)) {
+        const [bitsw_score, percent_div, percent_del, percent_ins, query_chr, query_begin, query_end, query_remaining, orientation, matching_repeat_name, matching_repeat_class, matching_repeat_begin, matching_repeat_end, matching_repeat_remaining, repeat_id,] = description.trim().split(' ');
+        return {
+            bitsw_score,
+            percent_div,
+            percent_del,
+            percent_ins,
+            query_chr,
+            query_begin,
+            query_end,
+            query_remaining,
+            orientation,
+            matching_repeat_name,
+            matching_repeat_class,
+            matching_repeat_begin,
+            matching_repeat_end,
+            matching_repeat_remaining,
+            repeat_id,
+        };
+    }
+    return { description };
+}
+export function generateRepeatMaskerFeature({ uniqueId, refName, start, end, description, ...rest }) {
+    return {
+        ...rest,
+        ...makeRepeatTrackDescription(description),
+        uniqueId,
+        refName,
+        start,
+        end,
+    };
+}

package/esm/generateUcscTranscript.d.ts

@@ -0,0 +1,15 @@
+import { MinimalFeature, TranscriptFeat } from './types';
+export declare function isUcscTranscript({ thickStart, blockCount, strand, }: {
+    thickStart?: number;
+    blockCount?: number;
+    strand?: number;
+}): boolean | 0 | undefined;
+export declare function generateUcscTranscript(data: TranscriptFeat): {
+    uniqueId: string;
+    strand: number;
+    type: string;
+    refName: string;
+    subfeatures: MinimalFeature[];
+    start: number;
+    end: number;
+};

package/esm/generateUcscTranscript.js

@@ -0,0 +1,97 @@
+export function isUcscTranscript({ thickStart, blockCount, strand, }) {
+    return thickStart && blockCount && strand !== 0;
+}
+export function generateUcscTranscript(data) {
+    const { strand = 0, chrom: _1, chromStart: _2, chromEnd: _3, chromStarts, blockStarts, blockSizes, uniqueId, ...rest } = data;
+    const { subfeatures: oldSubfeatures, thickStart, thickEnd, blockCount, refName, ...rest2 } = rest;
+    const subfeatures = [];
+    const feats = oldSubfeatures
+        .filter(child => child.type === 'block')
+        .sort((a, b) => a.start - b.start);
+    for (const block of feats) {
+        const start = block.start;
+        const end = block.end;
+        if (thickStart >= end) {
+            // left-side UTR
+            subfeatures.push({
+                type: `${strand > 0 ? 'five' : 'three'}_prime_UTR`,
+                start,
+                end,
+                refName,
+            });
+        }
+        else if (thickStart > start && thickStart < end && thickEnd >= end) {
+            // UTR | CDS
+            subfeatures.push({
+                type: `${strand > 0 ? 'five' : 'three'}_prime_UTR`,
+                start,
+                end: thickStart,
+                refName,
+            }, {
+                type: 'CDS',
+                start: thickStart,
+                end,
+                refName,
+            });
+        }
+        else if (thickStart <= start && thickEnd >= end) {
+            // CDS
+            subfeatures.push({
+                type: 'CDS',
+                start,
+                end,
+                refName,
+            });
+        }
+        else if (thickStart > start && thickStart < end && thickEnd < end) {
+            // UTR | CDS | UTR
+            subfeatures.push({
+                type: `${strand > 0 ? 'five' : 'three'}_prime_UTR`,
+                start,
+                end: thickStart,
+                refName,
+            }, {
+                type: 'CDS',
+                start: thickStart,
+                end: thickEnd,
+                refName,
+            }, {
+                type: `${strand > 0 ? 'three' : 'five'}_prime_UTR`,
+                start: thickEnd,
+                end,
+                refName,
+            });
+        }
+        else if (thickStart <= start && thickEnd > start && thickEnd < end) {
+            // CDS | UTR
+            subfeatures.push({
+                type: 'CDS',
+                start,
+                end: thickEnd,
+                refName,
+            }, {
+                type: `${strand > 0 ? 'three' : 'five'}_prime_UTR`,
+                start: thickEnd,
+                end,
+                refName,
+            });
+        }
+        else if (thickEnd <= start) {
+            // right-side UTR
+            subfeatures.push({
+                type: `${strand > 0 ? 'three' : 'five'}_prime_UTR`,
+                start,
+                end,
+                refName,
+            });
+        }
+    }
+    return {
+        ...rest2,
+        uniqueId,
+        strand,
+        type: 'mRNA',
+        refName,
+        subfeatures,
+    };
+}

package/esm/types.d.ts

@@ -0,0 +1,18 @@
+export interface MinimalFeature {
+    type: string;
+    start: number;
+    end: number;
+    refName: string;
+    [key: string]: unknown;
+}
+export interface TranscriptFeat extends MinimalFeature {
+    uniqueId: string;
+    thickStart: number;
+    thickEnd: number;
+    blockCount: number;
+    blockSizes: number[];
+    chromStarts: number[];
+    refName: string;
+    strand?: number;
+    subfeatures: MinimalFeature[];
+}

package/esm/types.js

@@ -0,0 +1 @@
+export {};

package/esm/util.d.ts

@@ -1,37 +1,12 @@
 import BED from '@gmod/bed';
-import { SimpleFeature } from '@jbrowse/core/util';
-export interface MinimalFeature {
-    type: string;
-    start: number;
-    end: number;
-    refName: string;
-    [key: string]: unknown;
-}
-export interface TranscriptFeat extends MinimalFeature {
-    thickStart: number;
-    thickEnd: number;
-    blockCount: number;
-    blockSizes: number[];
-    chromStarts: number[];
-    refName: string;
-    strand?: number;
-    subfeatures: MinimalFeature[];
-}
-export declare function ucscProcessedTranscript(feature: TranscriptFeat): TranscriptFeat | {
-    strand: number;
-    type: string;
-    refName: string;
-    subfeatures: MinimalFeature[];
-    start: number;
-    end: number;
-};
+import { SimpleFeatureSerialized } from '@jbrowse/core/util';
 export declare function makeBlocks({ start, uniqueId, refName, chromStarts, blockCount, blockSizes, blockStarts, }: {
     blockCount: number;
     start: number;
     uniqueId: string;
     refName: string;
     chromStarts?: number[];
-    blockSizes: number[];
+    blockSizes?: number[];
     blockStarts?: number[];
 }): {
     uniqueId: string;
@@ -40,9 +15,24 @@ export declare function makeBlocks({ start, uniqueId, refName, chromStarts, bloc
     refName: string;
     type: string;
 }[];
-export declare function featureData(line: string, colRef: number, colStart: number, colEnd: number, scoreColumn: string, parser: BED, uniqueId: string, names?: string[]): SimpleFeature;
-export declare function isUcscProcessedTranscript(f: {
-    thickStart?: number;
-    blockCount?: number;
-    strand?: number;
-}): boolean | 0 | undefined;
+export declare function featureData({ line, colRef, colStart, colEnd, scoreColumn, parser, uniqueId, names, }: {
+    line: string;
+    colRef: number;
+    colStart: number;
+    colEnd: number;
+    scoreColumn: string;
+    parser: BED;
+    uniqueId: string;
+    names?: string[];
+}): SimpleFeatureSerialized;
+export declare function featureData2({ line, refName, start, end, parser, uniqueId, scoreColumn, names, }: {
+    line: string;
+    refName: string;
+    start: number;
+    end: number;
+    parser: BED;
+    uniqueId: string;
+    scoreColumn: string;
+    names?: string[];
+}): SimpleFeatureSerialized;
+export declare function arrayify(f?: string | number[]): number[] | undefined;