@jbrowse/plugin-alignments 1.7.9 → 2.0.0

package/esm/BamAdapter/BamAdapter.js

@@ -0,0 +1,173 @@
+import { BamFile } from '@gmod/bam';
+import { BaseFeatureDataAdapter, } from '@jbrowse/core/data_adapters/BaseAdapter';
+import { bytesForRegions, updateStatus } from '@jbrowse/core/util';
+import { openLocation } from '@jbrowse/core/util/io';
+import { ObservableCreate } from '@jbrowse/core/util/rxjs';
+import { toArray } from 'rxjs/operators';
+import { readConfObject } from '@jbrowse/core/configuration';
+import BamSlightlyLazyFeature from './BamSlightlyLazyFeature';
+export default class BamAdapter extends BaseFeatureDataAdapter {
+    // derived classes may not use the same configuration so a custom
+    // configure method allows derived classes to override this behavior
+    async configure() {
+        if (!this.configured) {
+            const bamLocation = readConfObject(this.config, 'bamLocation');
+            const location = readConfObject(this.config, ['index', 'location']);
+            const indexType = readConfObject(this.config, ['index', 'indexType']);
+            const bam = new BamFile({
+                bamFilehandle: openLocation(bamLocation, this.pluginManager),
+                csiFilehandle: indexType === 'CSI'
+                    ? openLocation(location, this.pluginManager)
+                    : undefined,
+                baiFilehandle: indexType !== 'CSI'
+                    ? openLocation(location, this.pluginManager)
+                    : undefined,
+                // chunkSizeLimit and fetchSizeLimit are more troublesome than
+                // helpful, and have given overly large values on the ultra long
+                // nanopore reads even with 500MB limits, so disabled with infinity
+                chunkSizeLimit: Infinity,
+                fetchSizeLimit: Infinity,
+                yieldThreadTime: Infinity,
+            });
+            const adapterConfig = readConfObject(this.config, 'sequenceAdapter');
+            if (adapterConfig && this.getSubAdapter) {
+                this.configured = this.getSubAdapter(adapterConfig).then(({ dataAdapter }) => ({
+                    bam,
+                    sequenceAdapter: dataAdapter,
+                }));
+            }
+            else {
+                this.configured = Promise.resolve({ bam });
+            }
+        }
+        return this.configured;
+    }
+    async getHeader(opts) {
+        const { bam } = await this.configure();
+        return bam.getHeaderText(opts);
+    }
+    async setupPre(opts) {
+        const { statusCallback = () => { } } = opts || {};
+        const { bam } = await this.configure();
+        this.samHeader = await updateStatus('Downloading index', statusCallback, async () => {
+            const samHeader = await bam.getHeader(opts);
+            // use the @SQ lines in the header to figure out the
+            // mapping between ref ref ID numbers and names
+            const idToName = [];
+            const nameToId = {};
+            samHeader
+                .filter(l => l.tag === 'SQ')
+                .forEach((sqLine, refId) => {
+                sqLine.data.forEach(item => {
+                    if (item.tag === 'SN') {
+                        // this is the ref name
+                        const refName = item.value;
+                        nameToId[refName] = refId;
+                        idToName[refId] = refName;
+                    }
+                });
+            });
+            return { idToName, nameToId };
+        });
+        return this.samHeader;
+    }
+    async setup(opts) {
+        if (!this.setupP) {
+            this.setupP = this.setupPre(opts).catch(e => {
+                this.setupP = undefined;
+                throw e;
+            });
+        }
+        return this.setupP;
+    }
+    async getRefNames(opts) {
+        const { idToName } = await this.setup(opts);
+        return idToName;
+    }
+    async seqFetch(refName, start, end) {
+        const { sequenceAdapter } = await this.configure();
+        const refSeqStore = sequenceAdapter;
+        if (!refSeqStore) {
+            return undefined;
+        }
+        if (!refName) {
+            return undefined;
+        }
+        const features = refSeqStore.getFeatures({
+            refName,
+            start,
+            end,
+            assemblyName: '',
+        });
+        const seqChunks = await features.pipe(toArray()).toPromise();
+        let sequence = '';
+        seqChunks
+            .sort((a, b) => a.get('start') - b.get('start'))
+            .forEach(chunk => {
+            const chunkStart = chunk.get('start');
+            const chunkEnd = chunk.get('end');
+            const trimStart = Math.max(start - chunkStart, 0);
+            const trimEnd = Math.min(end - chunkStart, chunkEnd - chunkStart);
+            const trimLength = trimEnd - trimStart;
+            const chunkSeq = chunk.get('seq') || chunk.get('residues');
+            sequence += chunkSeq.substr(trimStart, trimLength);
+        });
+        if (sequence.length !== end - start) {
+            throw new Error(`sequence fetch failed: fetching ${refName}:${(start - 1).toLocaleString()}-${end.toLocaleString()} returned ${sequence.length.toLocaleString()} bases, but should have returned ${(end - start).toLocaleString()}`);
+        }
+        return sequence;
+    }
+    getFeatures(region, opts) {
+        const { refName, start, end, originalRefName } = region;
+        const { signal, filterBy, statusCallback = () => { } } = opts || {};
+        return ObservableCreate(async (observer) => {
+            const { bam } = await this.configure();
+            await this.setup(opts);
+            statusCallback('Downloading alignments');
+            const records = await bam.getRecordsForRange(refName, start, end, opts);
+            const { flagInclude = 0, flagExclude = 0, tagFilter, readName, } = filterBy || {};
+            for (const record of records) {
+                let ref;
+                if (!record.get('MD')) {
+                    ref = await this.seqFetch(originalRefName || refName, record.get('start'), record.get('end'));
+                }
+                const flags = record.flags;
+                if (!((flags & flagInclude) === flagInclude && !(flags & flagExclude))) {
+                    continue;
+                }
+                if (tagFilter) {
+                    const val = record.get(tagFilter.tag);
+                    if (!(val === '*' ? val !== undefined : val === tagFilter.value)) {
+                        continue;
+                    }
+                }
+                if (readName && record.get('name') !== readName) {
+                    continue;
+                }
+                observer.next(new BamSlightlyLazyFeature(record, this, ref));
+            }
+            statusCallback('');
+            observer.complete();
+        }, signal);
+    }
+    async estimateRegionsStats(regions, opts) {
+        const { bam } = await this.configure();
+        // this is a method to avoid calling on htsget adapters
+        // @ts-ignore
+        if (bam.index.filehandle !== '?') {
+            const bytes = await bytesForRegions(regions, bam);
+            const fetchSizeLimit = readConfObject(this.config, 'fetchSizeLimit');
+            return { bytes, fetchSizeLimit };
+        }
+        else {
+            return super.estimateRegionsStats(regions, opts);
+        }
+    }
+    freeResources( /* { region } */) { }
+    // depends on setup being called before the BAM constructor
+    refIdToName(refId) {
+        var _a;
+        return (_a = this.samHeader) === null || _a === void 0 ? void 0 : _a.idToName[refId];
+    }
+}
+//# sourceMappingURL=BamAdapter.js.map

package/esm/BamAdapter/BamAdapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"BamAdapter.js","sourceRoot":"","sources":["../../src/BamAdapter/BamAdapter.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,OAAO,EAAE,MAAM,WAAW,CAAA;AACnC,OAAO,EACL,sBAAsB,GAEvB,MAAM,yCAAyC,CAAA;AAEhD,OAAO,EAAE,eAAe,EAAE,YAAY,EAAE,MAAM,oBAAoB,CAAA;AAClE,OAAO,EAAE,YAAY,EAAE,MAAM,uBAAuB,CAAA;AACpD,OAAO,EAAE,gBAAgB,EAAE,MAAM,yBAAyB,CAAA;AAE1D,OAAO,EAAE,OAAO,EAAE,MAAM,gBAAgB,CAAA;AACxC,OAAO,EAAE,cAAc,EAAE,MAAM,6BAA6B,CAAA;AAC5D,OAAO,sBAAsB,MAAM,0BAA0B,CAAA;AAO7D,MAAM,CAAC,OAAO,OAAO,UAAW,SAAQ,sBAAsB;IAU5D,iEAAiE;IACjE,oEAAoE;IAC1D,KAAK,CAAC,SAAS;QACvB,IAAI,CAAC,IAAI,CAAC,UAAU,EAAE;YACpB,MAAM,WAAW,GAAG,cAAc,CAAC,IAAI,CAAC,MAAM,EAAE,aAAa,CAAC,CAAA;YAC9D,MAAM,QAAQ,GAAG,cAAc,CAAC,IAAI,CAAC,MAAM,EAAE,CAAC,OAAO,EAAE,UAAU,CAAC,CAAC,CAAA;YACnE,MAAM,SAAS,GAAG,cAAc,CAAC,IAAI,CAAC,MAAM,EAAE,CAAC,OAAO,EAAE,WAAW,CAAC,CAAC,CAAA;YACrE,MAAM,GAAG,GAAG,IAAI,OAAO,CAAC;gBACtB,aAAa,EAAE,YAAY,CAAC,WAAW,EAAE,IAAI,CAAC,aAAa,CAAC;gBAC5D,aAAa,EACX,SAAS,KAAK,KAAK;oBACjB,CAAC,CAAC,YAAY,CAAC,QAAQ,EAAE,IAAI,CAAC,aAAa,CAAC;oBAC5C,CAAC,CAAC,SAAS;gBACf,aAAa,EACX,SAAS,KAAK,KAAK;oBACjB,CAAC,CAAC,YAAY,CAAC,QAAQ,EAAE,IAAI,CAAC,aAAa,CAAC;oBAC5C,CAAC,CAAC,SAAS;gBAEf,8DAA8D;gBAC9D,gEAAgE;gBAChE,mEAAmE;gBACnE,cAAc,EAAE,QAAQ;gBACxB,cAAc,EAAE,QAAQ;gBACxB,eAAe,EAAE,QAAQ;aAC1B,CAAC,CAAA;YAEF,MAAM,aAAa,GAAG,cAAc,CAAC,IAAI,CAAC,MAAM,EAAE,iBAAiB,CAAC,CAAA;YACpE,IAAI,aAAa,IAAI,IAAI,CAAC,aAAa,EAAE;gBACvC,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC,aAAa,CAAC,aAAa,CAAC,CAAC,IAAI,CACtD,CAAC,EAAE,WAAW,EAAE,EAAE,EAAE,CAAC,CAAC;oBACpB,GAAG;oBACH,eAAe,EAAE,WAAqC;iBACvD,CAAC,CACH,CAAA;aACF;iBAAM;gBACL,IAAI,CAAC,UAAU,GAAG,OAAO,CAAC,OAAO,CAAC,EAAE,GAAG,EAAE,CAAC,CAAA;aAC3C;SACF;QACD,OAAO,IAAI,CAAC,UAAU,CAAA;IACxB,CAAC;IAED,KAAK,CAAC,SAAS,CAAC,IAAkB;QAChC,MAAM,EAAE,GAAG,EAAE,GAAG,MAAM,IAAI,CAAC,SAAS,EAAE,CAAA;QACtC,OAAO,GAAG,CAAC,aAAa,CAAC,IAAI,CAAC,CAAA;IAChC,CAAC;IAEO,KAAK,CAAC,QAAQ,CAAC,IAAkB;QACvC,MAAM,EAAE,cAAc,GAAG,GAAG,EAAE,GAAE,CAAC,EAAE,GAAG,IAAI,IAAI,EAAE,CAAA;QAChD,MAAM,EAAE,GAAG,EAAE,GAAG,MAAM,IAAI,CAAC,SAAS,EAAE,CAAA;QACtC,IAAI,CAAC,SAAS,GAAG,MAAM,YAAY,CACjC,mBAAmB,EACnB,cAAc,EACd,KAAK,IAAI,EAAE;YACT,MAAM,SAAS,GAAG,MAAM,GAAG,CAAC,SAAS,CAAC,IAAI,CAAC,CAAA;YAE3C,oDAAoD;YACpD,+CAA+C;YAC/C,MAAM,QAAQ,GAAa,EAAE,CAAA;YAC7B,MAAM,QAAQ,GAA2B,EAAE,CAAA;YAC3C,SAAS;iBACN,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,GAAG,KAAK,IAAI,CAAC;iBAC3B,OAAO,CAAC,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE;gBACzB,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE;oBACzB,IAAI,IAAI,CAAC,GAAG,KAAK,IAAI,EAAE;wBACrB,uBAAuB;wBACvB,MAAM,OAAO,GAAG,IAAI,CAAC,KAAK,CAAA;wBAC1B,QAAQ,CAAC,OAAO,CAAC,GAAG,KAAK,CAAA;wBACzB,QAAQ,CAAC,KAAK,CAAC,GAAG,OAAO,CAAA;qBAC1B;gBACH,CAAC,CAAC,CAAA;YACJ,CAAC,CAAC,CAAA;YAEJ,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,CAAA;QAC/B,CAAC,CACF,CAAA;QACD,OAAO,IAAI,CAAC,SAAS,CAAA;IACvB,CAAC;IAEO,KAAK,CAAC,KAAK,CAAC,IAAkB;QACpC,IAAI,CAAC,IAAI,CAAC,MAAM,EAAE;YAChB,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE;gBAC1C,IAAI,CAAC,MAAM,GAAG,SAAS,CAAA;gBACvB,MAAM,CAAC,CAAA;YACT,CAAC,CAAC,CAAA;SACH;QACD,OAAO,IAAI,CAAC,MAAM,CAAA;IACpB,CAAC;IAED,KAAK,CAAC,WAAW,CAAC,IAAkB;QAClC,MAAM,EAAE,QAAQ,EAAE,GAAG,MAAM,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAA;QAC3C,OAAO,QAAQ,CAAA;IACjB,CAAC;IAEO,KAAK,CAAC,QAAQ,CAAC,OAAe,EAAE,KAAa,EAAE,GAAW;QAChE,MAAM,EAAE,eAAe,EAAE,GAAG,MAAM,IAAI,CAAC,SAAS,EAAE,CAAA;QAClD,MAAM,WAAW,GAAG,eAAe,CAAA;QACnC,IAAI,CAAC,WAAW,EAAE;YAChB,OAAO,SAAS,CAAA;SACjB;QACD,IAAI,CAAC,OAAO,EAAE;YACZ,OAAO,SAAS,CAAA;SACjB;QAED,MAAM,QAAQ,GAAG,WAAW,CAAC,WAAW,CAAC;YACvC,OAAO;YACP,KAAK;YACL,GAAG;YACH,YAAY,EAAE,EAAE;SACjB,CAAC,CAAA;QAEF,MAAM,SAAS,GAAG,MAAM,QAAQ,CAAC,IAAI,CAAC,OAAO,EAAE,CAAC,CAAC,SAAS,EAAE,CAAA;QAE5D,IAAI,QAAQ,GAAG,EAAE,CAAA;QACjB,SAAS;aACN,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC;aAC/C,OAAO,CAAC,KAAK,CAAC,EAAE;YACf,MAAM,UAAU,GAAG,KAAK,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;YACrC,MAAM,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,CAAA;YACjC,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,KAAK,GAAG,UAAU,EAAE,CAAC,CAAC,CAAA;YACjD,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,GAAG,UAAU,EAAE,QAAQ,GAAG,UAAU,CAAC,CAAA;YACjE,MAAM,UAAU,GAAG,OAAO,GAAG,SAAS,CAAA;YACtC,MAAM,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,KAAK,CAAC,GAAG,CAAC,UAAU,CAAC,CAAA;YAC1D,QAAQ,IAAI,QAAQ,CAAC,MAAM,CAAC,SAAS,EAAE,UAAU,CAAC,CAAA;QACpD,CAAC,CAAC,CAAA;QAEJ,IAAI,QAAQ,CAAC,MAAM,KAAK,GAAG,GAAG,KAAK,EAAE;YACnC,MAAM,IAAI,KAAK,CACb,mCAAmC,OAAO,IAAI,CAC5C,KAAK,GAAG,CAAC,CACV,CAAC,cAAc,EAAE,IAAI,GAAG,CAAC,cAAc,EAAE,aAAa,QAAQ,CAAC,MAAM,CAAC,cAAc,EAAE,oCAAoC,CACzH,GAAG,GAAG,KAAK,CACZ,CAAC,cAAc,EAAE,EAAE,CACrB,CAAA;SACF;QACD,OAAO,QAAQ,CAAA;IACjB,CAAC;IAED,WAAW,CACT,MAA6C,EAC7C,IAOC;QAED,MAAM,EAAE,OAAO,EAAE,KAAK,EAAE,GAAG,EAAE,eAAe,EAAE,GAAG,MAAM,CAAA;QACvD,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,cAAc,GAAG,GAAG,EAAE,GAAE,CAAC,EAAE,GAAG,IAAI,IAAI,EAAE,CAAA;QAClE,OAAO,gBAAgB,CAAU,KAAK,EAAC,QAAQ,EAAC,EAAE;YAChD,MAAM,EAAE,GAAG,EAAE,GAAG,MAAM,IAAI,CAAC,SAAS,EAAE,CAAA;YACtC,MAAM,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAA;YACtB,cAAc,CAAC,wBAAwB,CAAC,CAAA;YACxC,MAAM,OAAO,GAAG,MAAM,GAAG,CAAC,kBAAkB,CAAC,OAAO,EAAE,KAAK,EAAE,GAAG,EAAE,IAAI,CAAC,CAAA;YAEvE,MAAM,EACJ,WAAW,GAAG,CAAC,EACf,WAAW,GAAG,CAAC,EACf,SAAS,EACT,QAAQ,GACT,GAAG,QAAQ,IAAI,EAAE,CAAA;YAElB,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE;gBAC5B,IAAI,GAAuB,CAAA;gBAC3B,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE;oBACrB,GAAG,GAAG,MAAM,IAAI,CAAC,QAAQ,CACvB,eAAe,IAAI,OAAO,EAC1B,MAAM,CAAC,GAAG,CAAC,OAAO,CAAC,EACnB,MAAM,CAAC,GAAG,CAAC,KAAK,CAAC,CAClB,CAAA;iBACF;gBAED,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAA;gBAC1B,IACE,CAAC,CAAC,CAAC,KAAK,GAAG,WAAW,CAAC,KAAK,WAAW,IAAI,CAAC,CAAC,KAAK,GAAG,WAAW,CAAC,CAAC,EAClE;oBACA,SAAQ;iBACT;gBAED,IAAI,SAAS,EAAE;oBACb,MAAM,GAAG,GAAG,MAAM,CAAC,GAAG,CAAC,SAAS,CAAC,GAAG,CAAC,CAAA;oBACrC,IAAI,CAAC,CAAC,GAAG,KAAK,GAAG,CAAC,CAAC,CAAC,GAAG,KAAK,SAAS,CAAC,CAAC,CAAC,GAAG,KAAK,SAAS,CAAC,KAAK,CAAC,EAAE;wBAChE,SAAQ;qBACT;iBACF;gBAED,IAAI,QAAQ,IAAI,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,KAAK,QAAQ,EAAE;oBAC/C,SAAQ;iBACT;gBAED,QAAQ,CAAC,IAAI,CAAC,IAAI,sBAAsB,CAAC,MAAM,EAAE,IAAI,EAAE,GAAG,CAAC,CAAC,CAAA;aAC7D;YACD,cAAc,CAAC,EAAE,CAAC,CAAA;YAClB,QAAQ,CAAC,QAAQ,EAAE,CAAA;QACrB,CAAC,EAAE,MAAM,CAAC,CAAA;IACZ,CAAC;IAED,KAAK,CAAC,oBAAoB,CAAC,OAAiB,EAAE,IAAkB;QAC9D,MAAM,EAAE,GAAG,EAAE,GAAG,MAAM,IAAI,CAAC,SAAS,EAAE,CAAA;QACtC,uDAAuD;QACvD,aAAa;QACb,IAAI,GAAG,CAAC,KAAK,CAAC,UAAU,KAAK,GAAG,EAAE;YAChC,MAAM,KAAK,GAAG,MAAM,eAAe,CAAC,OAAO,EAAE,GAAG,CAAC,CAAA;YACjD,MAAM,cAAc,GAAG,cAAc,CAAC,IAAI,CAAC,MAAM,EAAE,gBAAgB,CAAC,CAAA;YACpE,OAAO,EAAE,KAAK,EAAE,cAAc,EAAE,CAAA;SACjC;aAAM;YACL,OAAO,KAAK,CAAC,oBAAoB,CAAC,OAAO,EAAE,IAAI,CAAC,CAAA;SACjD;IACH,CAAC;IAED,aAAa,EAAC,gBAAgB,IAAS,CAAC;IAExC,2DAA2D;IAC3D,WAAW,CAAC,KAAa;;QACvB,OAAO,MAAA,IAAI,CAAC,SAAS,0CAAE,QAAQ,CAAC,KAAK,CAAC,CAAA;IACxC,CAAC;CACF"}

package/esm/BamAdapter/BamSlightlyLazyFeature.d.ts

@@ -0,0 +1,33 @@
+/// <reference types="node" />
+import { Feature, SimpleFeatureSerialized } from '@jbrowse/core/util/simpleFeature';
+import { BamRecord } from '@gmod/bam';
+import BamAdapter from './BamAdapter';
+export default class BamSlightlyLazyFeature implements Feature {
+    private record;
+    private adapter;
+    private ref?;
+    constructor(record: BamRecord, adapter: BamAdapter, ref?: string | undefined);
+    _get_name(): any;
+    _get_type(): string;
+    _get_score(): number;
+    _get_flags(): string;
+    _get_strand(): number;
+    _get_pair_orientation(): string | null | undefined;
+    _get_next_seq_id(): number;
+    _get_seq_id(): number;
+    _get_next_refName(): string | undefined;
+    _get_next_segment_position(): string | undefined;
+    _get_seq(): string;
+    qualRaw(): Buffer | undefined;
+    set(): void;
+    tags(): string[];
+    id(): string;
+    get(field: string): any;
+    _get_refName(): string | undefined;
+    parent(): undefined;
+    children(): undefined;
+    pairedFeature(): boolean;
+    toJSON(): SimpleFeatureSerialized;
+    _get_mismatches(): import("./MismatchParser").Mismatch[];
+    _get_clipPos(): number;
+}

package/esm/BamAdapter/BamSlightlyLazyFeature.js

@@ -0,0 +1,107 @@
+import { getMismatches } from './MismatchParser';
+export default class BamSlightlyLazyFeature {
+    // uses parameter properties to automatically create fields on the class
+    // https://www.typescriptlang.org/docs/handbook/classes.html#parameter-properties
+    constructor(record, adapter, ref) {
+        this.record = record;
+        this.adapter = adapter;
+        this.ref = ref;
+    }
+    _get_name() {
+        return this.record.get('name');
+    }
+    _get_type() {
+        return 'match';
+    }
+    _get_score() {
+        return this.record.get('mq');
+    }
+    _get_flags() {
+        return this.record.flags;
+    }
+    _get_strand() {
+        return this.record.isReverseComplemented() ? -1 : 1;
+    }
+    _get_pair_orientation() {
+        return this.record.isPaired() ? this.record.getPairOrientation() : undefined;
+    }
+    _get_next_seq_id() {
+        return this.record._next_refid();
+    }
+    _get_seq_id() {
+        // @ts-ignore
+        return this.record._refID;
+    }
+    _get_next_refName() {
+        return this.adapter.refIdToName(this.record._next_refid());
+    }
+    _get_next_segment_position() {
+        const { record, adapter } = this;
+        return record.isPaired()
+            ? `${adapter.refIdToName(record._next_refid())}:${record._next_pos() + 1}`
+            : undefined;
+    }
+    _get_seq() {
+        return this.record.getReadBases();
+    }
+    qualRaw() {
+        return this.record.qualRaw();
+    }
+    set() { }
+    tags() {
+        const properties = Object.getOwnPropertyNames(BamSlightlyLazyFeature.prototype);
+        return [
+            ...new Set(properties
+                .filter(prop => prop.startsWith('_get_') &&
+                prop !== '_get_mismatches' &&
+                prop !== '_get_tags' &&
+                prop !== '_get_next_seq_id' &&
+                prop !== '_get_seq_id')
+                .map(methodName => methodName.replace('_get_', ''))
+                .concat(this.record._tags())),
+        ];
+    }
+    id() {
+        return `${this.adapter.id}-${this.record.id()}`;
+    }
+    // eslint-disable-next-line @typescript-eslint/no-explicit-any
+    get(field) {
+        const methodName = `_get_${field}`;
+        // @ts-ignore
+        if (this[methodName]) {
+            // @ts-ignore
+            return this[methodName]();
+        }
+        return this.record.get(field);
+    }
+    _get_refName() {
+        return this.adapter.refIdToName(this.record.seq_id());
+    }
+    parent() {
+        return undefined;
+    }
+    children() {
+        return undefined;
+    }
+    pairedFeature() {
+        return false;
+    }
+    toJSON() {
+        return {
+            ...Object.fromEntries(this.tags()
+                .map(t => [t, this.get(t)])
+                .filter(elt => elt[1] !== undefined)),
+            uniqueId: this.id(),
+        };
+    }
+    _get_mismatches() {
+        return getMismatches(this.get('CIGAR'), this.get('MD'), this.get('seq'), this.ref, this.qualRaw());
+    }
+    _get_clipPos() {
+        const cigar = this.get('CIGAR') || '';
+        return this.get('strand') === -1
+            ? +(cigar.match(/(\d+)[SH]$/) || [])[1] || 0
+            : +(cigar.match(/^(\d+)([SH])/) || [])[1] || 0;
+    }
+}
+//# sourceMappingURL=BamSlightlyLazyFeature.js.map

package/esm/BamAdapter/BamSlightlyLazyFeature.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"BamSlightlyLazyFeature.js","sourceRoot":"","sources":["../../src/BamAdapter/BamSlightlyLazyFeature.ts"],"names":[],"mappings":"AAMA,OAAO,EAAE,aAAa,EAAE,MAAM,kBAAkB,CAAA;AAIhD,MAAM,CAAC,OAAO,OAAO,sBAAsB;IACzC,wEAAwE;IACxE,iFAAiF;IACjF,YACU,MAAiB,EACjB,OAAmB,EACnB,GAAY;QAFZ,WAAM,GAAN,MAAM,CAAW;QACjB,YAAO,GAAP,OAAO,CAAY;QACnB,QAAG,GAAH,GAAG,CAAS;IACnB,CAAC;IAEJ,SAAS;QACP,OAAO,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,CAAA;IAChC,CAAC;IAED,SAAS;QACP,OAAO,OAAO,CAAA;IAChB,CAAC;IAED,UAAU;QACR,OAAO,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;IAC9B,CAAC;IAED,UAAU;QACR,OAAO,IAAI,CAAC,MAAM,CAAC,KAAK,CAAA;IAC1B,CAAC;IAED,WAAW;QACT,OAAO,IAAI,CAAC,MAAM,CAAC,qBAAqB,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;IACrD,CAAC;IAED,qBAAqB;QACnB,OAAO,IAAI,CAAC,MAAM,CAAC,QAAQ,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,kBAAkB,EAAE,CAAC,CAAC,CAAC,SAAS,CAAA;IAC9E,CAAC;IAED,gBAAgB;QACd,OAAO,IAAI,CAAC,MAAM,CAAC,WAAW,EAAE,CAAA;IAClC,CAAC;IAED,WAAW;QACT,aAAa;QACb,OAAO,IAAI,CAAC,MAAM,CAAC,MAAM,CAAA;IAC3B,CAAC;IAED,iBAAiB;QACf,OAAO,IAAI,CAAC,OAAO,CAAC,WAAW,CAAC,IAAI,CAAC,MAAM,CAAC,WAAW,EAAE,CAAC,CAAA;IAC5D,CAAC;IAED,0BAA0B;QACxB,MAAM,EAAE,MAAM,EAAE,OAAO,EAAE,GAAG,IAAI,CAAA;QAChC,OAAO,MAAM,CAAC,QAAQ,EAAE;YACtB,CAAC,CAAC,GAAG,OAAO,CAAC,WAAW,CAAC,MAAM,CAAC,WAAW,EAAE,CAAC,IAAI,MAAM,CAAC,SAAS,EAAE,GAAG,CAAC,EAAE;YAC1E,CAAC,CAAC,SAAS,CAAA;IACf,CAAC;IAED,QAAQ;QACN,OAAO,IAAI,CAAC,MAAM,CAAC,YAAY,EAAE,CAAA;IACnC,CAAC;IAED,OAAO;QACL,OAAO,IAAI,CAAC,MAAM,CAAC,OAAO,EAAE,CAAA;IAC9B,CAAC;IAED,GAAG,KAAI,CAAC;IAER,IAAI;QACF,MAAM,UAAU,GAAG,MAAM,CAAC,mBAAmB,CAC3C,sBAAsB,CAAC,SAAS,CACjC,CAAA;QAED,OAAO;YACL,GAAG,IAAI,GAAG,CACR,UAAU;iBACP,MAAM,CACL,IAAI,CAAC,EAAE,CACL,IAAI,CAAC,UAAU,CAAC,OAAO,CAAC;gBACxB,IAAI,KAAK,iBAAiB;gBAC1B,IAAI,KAAK,WAAW;gBACpB,IAAI,KAAK,kBAAkB;gBAC3B,IAAI,KAAK,aAAa,CACzB;iBACA,GAAG,CAAC,UAAU,CAAC,EAAE,CAAC,UAAU,CAAC,OAAO,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC;iBAClD,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,KAAK,EAAE,CAAC,CAC/B;SACF,CAAA;IACH,CAAC;IAED,EAAE;QACA,OAAO,GAAG,IAAI,CAAC,OAAO,CAAC,EAAE,IAAI,IAAI,CAAC,MAAM,CAAC,EAAE,EAAE,EAAE,CAAA;IACjD,CAAC;IAED,8DAA8D;IAC9D,GAAG,CAAC,KAAa;QACf,MAAM,UAAU,GAAG,QAAQ,KAAK,EAAE,CAAA;QAClC,aAAa;QACb,IAAI,IAAI,CAAC,UAAU,CAAC,EAAE;YACpB,aAAa;YACb,OAAO,IAAI,CAAC,UAAU,CAAC,EAAE,CAAA;SAC1B;QACD,OAAO,IAAI,CAAC,MAAM,CAAC,GAAG,CAAC,KAAK,CAAC,CAAA;IAC/B,CAAC;IAED,YAAY;QACV,OAAO,IAAI,CAAC,OAAO,CAAC,WAAW,CAAC,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC,CAAA;IACvD,CAAC;IAED,MAAM;QACJ,OAAO,SAAS,CAAA;IAClB,CAAC;IAED,QAAQ;QACN,OAAO,SAAS,CAAA;IAClB,CAAC;IAED,aAAa;QACX,OAAO,KAAK,CAAA;IACd,CAAC;IAED,MAAM;QACJ,OAAO;YACL,GAAG,MAAM,CAAC,WAAW,CACnB,IAAI,CAAC,IAAI,EAAE;iBACR,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;iBAC1B,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC,CAAC,KAAK,SAAS,CAAC,CACvC;YACD,QAAQ,EAAE,IAAI,CAAC,EAAE,EAAE;SACpB,CAAA;IACH,CAAC;IAED,eAAe;QACb,OAAO,aAAa,CAClB,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,EACjB,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,EACd,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,EACf,IAAI,CAAC,GAAG,EACR,IAAI,CAAC,OAAO,EAAE,CACf,CAAA;IACH,CAAC;IAED,YAAY;QACV,MAAM,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,EAAE,CAAA;QACrC,OAAO,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC;YAC9B,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,KAAK,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;YAC5C,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,KAAK,CAAC,cAAc,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAA;IAClD,CAAC;CACF"}

package/esm/BamAdapter/MismatchParser.d.ts

@@ -0,0 +1,25 @@
+/// <reference types="node" />
+export interface Mismatch {
+    qual?: number;
+    start: number;
+    length: number;
+    type: string;
+    base: string;
+    altbase?: string;
+    seq?: string;
+    cliplen?: number;
+}
+export declare function parseCigar(cigar: string): string[];
+export declare function cigarToMismatches(ops: string[], seq: string, ref?: string, qual?: Buffer): Mismatch[];
+/**
+ * parse a SAM MD tag to find mismatching bases of the template versus the reference
+ * @returns array of mismatches and their positions
+ */
+export declare function mdToMismatches(mdstring: string, ops: string[], cigarMismatches: Mismatch[], seq: string, qual?: Buffer): Mismatch[];
+export declare function getMismatches(cigar: string, md: string, seq: string, ref?: string, qual?: Buffer): Mismatch[];
+export declare function getNextRefPos(cigarOps: string[], positions: number[]): Generator<number, void, unknown>;
+export declare function getModificationPositions(mm: string, fseq: string, fstrand: number): {
+    type: string;
+    positions: number[];
+}[];
+export declare function getModificationTypes(mm: string): string[];

package/esm/BamAdapter/MismatchParser.js

@@ -0,0 +1,294 @@
+import { revcom } from '@jbrowse/core/util';
+const mdRegex = new RegExp(/(\d+|\^[a-z]+|[a-z])/gi);
+const modificationRegex = new RegExp(/([A-Z])([-+])([^,.?]+)([.?])?/);
+export function parseCigar(cigar) {
+    return (cigar || '').split(/([MIDNSHPX=])/);
+}
+export function cigarToMismatches(ops, seq, ref, qual) {
+    let roffset = 0; // reference offset
+    let soffset = 0; // seq offset
+    const mismatches = [];
+    const hasRefAndSeq = ref && seq;
+    for (let i = 0; i < ops.length; i += 2) {
+        const len = +ops[i];
+        const op = ops[i + 1];
+        if (op === 'M' || op === '=' || op === 'E') {
+            if (hasRefAndSeq) {
+                for (let j = 0; j < len; j++) {
+                    if (
+                    // @ts-ignore in the full yarn build of the repo, this says that object is possibly undefined for some reason, ignored
+                    seq[soffset + j].toUpperCase() !== ref[roffset + j].toUpperCase()) {
+                        mismatches.push({
+                            start: roffset + j,
+                            type: 'mismatch',
+                            base: seq[soffset + j],
+                            altbase: ref[roffset + j],
+                            length: 1,
+                        });
+                    }
+                }
+            }
+            soffset += len;
+        }
+        if (op === 'I') {
+            mismatches.push({
+                start: roffset,
+                type: 'insertion',
+                base: `${len}`,
+                length: 0,
+            });
+            soffset += len;
+        }
+        else if (op === 'D') {
+            mismatches.push({
+                start: roffset,
+                type: 'deletion',
+                base: '*',
+                length: len,
+            });
+        }
+        else if (op === 'N') {
+            mismatches.push({
+                start: roffset,
+                type: 'skip',
+                base: 'N',
+                length: len,
+            });
+        }
+        else if (op === 'X') {
+            const r = seq.slice(soffset, soffset + len);
+            const q = (qual === null || qual === void 0 ? void 0 : qual.slice(soffset, soffset + len)) || [];
+            for (let j = 0; j < len; j++) {
+                mismatches.push({
+                    start: roffset + j,
+                    type: 'mismatch',
+                    base: r[j],
+                    qual: q[j],
+                    length: 1,
+                });
+            }
+            soffset += len;
+        }
+        else if (op === 'H') {
+            mismatches.push({
+                start: roffset,
+                type: 'hardclip',
+                base: `H${len}`,
+                cliplen: len,
+                length: 1,
+            });
+        }
+        else if (op === 'S') {
+            mismatches.push({
+                start: roffset,
+                type: 'softclip',
+                base: `S${len}`,
+                cliplen: len,
+                length: 1,
+            });
+            soffset += len;
+        }
+        if (op !== 'I' && op !== 'S' && op !== 'H') {
+            roffset += len;
+        }
+    }
+    return mismatches;
+}
+/**
+ * parse a SAM MD tag to find mismatching bases of the template versus the reference
+ * @returns array of mismatches and their positions
+ */
+export function mdToMismatches(mdstring, ops, cigarMismatches, seq, qual) {
+    const mismatchRecords = [];
+    let curr = { start: 0, base: '', length: 0, type: 'mismatch' };
+    const skips = cigarMismatches.filter(cigar => cigar.type === 'skip');
+    let lastCigar = 0;
+    let lastTemplateOffset = 0;
+    let lastRefOffset = 0;
+    let lastSkipPos = 0;
+    // convert a position on the reference sequence to a position
+    // on the template sequence, taking into account hard and soft
+    // clipping of reads
+    function nextRecord() {
+        mismatchRecords.push(curr);
+        // get a new mismatch record ready
+        curr = {
+            start: curr.start + curr.length,
+            length: 0,
+            base: '',
+            type: 'mismatch',
+        };
+    }
+    function getTemplateCoordLocal(refCoord) {
+        let templateOffset = lastTemplateOffset;
+        let refOffset = lastRefOffset;
+        for (let i = lastCigar; i < ops.length && refOffset <= refCoord; i += 2, lastCigar = i) {
+            const len = +ops[i];
+            const op = ops[i + 1];
+            if (op === 'S' || op === 'I') {
+                templateOffset += len;
+            }
+            else if (op === 'D' || op === 'P' || op === 'N') {
+                refOffset += len;
+            }
+            else if (op !== 'H') {
+                templateOffset += len;
+                refOffset += len;
+            }
+        }
+        lastTemplateOffset = templateOffset;
+        lastRefOffset = refOffset;
+        return templateOffset - (refOffset - refCoord);
+    }
+    // now actually parse the MD string
+    const md = mdstring.match(mdRegex) || [];
+    for (let i = 0; i < md.length; i++) {
+        const token = md[i];
+        const num = +token;
+        if (!Number.isNaN(num)) {
+            curr.start += num;
+        }
+        else if (token.startsWith('^')) {
+            curr.start += token.length - 1;
+        }
+        else {
+            // mismatch
+            for (let j = 0; j < token.length; j += 1) {
+                curr.length = 1;
+                while (lastSkipPos < skips.length) {
+                    const mismatch = skips[lastSkipPos];
+                    if (curr.start >= mismatch.start) {
+                        curr.start += mismatch.length;
+                        lastSkipPos++;
+                    }
+                    else {
+                        break;
+                    }
+                }
+                const s = getTemplateCoordLocal(curr.start);
+                curr.base = seq[s] || 'X';
+                const qualScore = qual === null || qual === void 0 ? void 0 : qual[s];
+                if (qualScore) {
+                    curr.qual = qualScore;
+                }
+                curr.altbase = token;
+                nextRecord();
+            }
+        }
+    }
+    return mismatchRecords;
+}
+export function getMismatches(cigar, md, seq, ref, qual) {
+    let mismatches = [];
+    const ops = parseCigar(cigar);
+    // parse the CIGAR tag if it has one
+    if (cigar) {
+        mismatches = mismatches.concat(cigarToMismatches(ops, seq, ref, qual));
+    }
+    // now let's look for CRAM or MD mismatches
+    if (md) {
+        mismatches = mismatches.concat(mdToMismatches(md, ops, mismatches, seq, qual));
+    }
+    return mismatches;
+}
+// get relative reference sequence positions for positions given relative to
+// the read sequence
+export function* getNextRefPos(cigarOps, positions) {
+    let readPos = 0;
+    let refPos = 0;
+    let currPos = 0;
+    for (let i = 0; i < cigarOps.length && currPos < positions.length; i += 2) {
+        const len = +cigarOps[i];
+        const op = cigarOps[i + 1];
+        if (op === 'S' || op === 'I') {
+            for (let i = 0; i < len && currPos < positions.length; i++) {
+                if (positions[currPos] === readPos + i) {
+                    currPos++;
+                }
+            }
+            readPos += len;
+        }
+        else if (op === 'D' || op === 'N') {
+            refPos += len;
+        }
+        else if (op === 'M' || op === 'X' || op === '=') {
+            for (let i = 0; i < len && currPos < positions.length; i++) {
+                if (positions[currPos] === readPos + i) {
+                    yield refPos + i;
+                    currPos++;
+                }
+            }
+            readPos += len;
+            refPos += len;
+        }
+    }
+}
+export function getModificationPositions(mm, fseq, fstrand) {
+    const seq = fstrand === -1 ? revcom(fseq) : fseq;
+    const mods = mm.split(';').filter(mod => !!mod);
+    const result = [];
+    for (let i = 0; i < mods.length; i++) {
+        const mod = mods[i];
+        const [basemod, ...skips] = mod.split(',');
+        // regexes based on parse_mm.pl from hts-specs
+        const matches = basemod.match(modificationRegex);
+        if (!matches) {
+            throw new Error('bad format for MM tag');
+        }
+        const [, base, strand, typestr] = matches;
+        // can be a multi e.g. C+mh for both meth (m) and hydroxymeth (h) so
+        // split, and they can also be chemical codes (ChEBI) e.g. C+16061
+        const types = typestr.split(/(\d+|.)/).filter(f => !!f);
+        if (strand === '-') {
+            console.warn('unsupported negative strand modifications');
+            // make sure to return a somewhat matching type even in this case
+            result.push({ type: 'unsupported', positions: [] });
+        }
+        // this logic also based on parse_mm.pl from hts-specs is that in the
+        // sequence of the read, if we have a modification type e.g. C+m;2 and a
+        // sequence ACGTACGTAC we skip the two instances of C and go to the last
+        // C
+        for (let j = 0; j < types.length; j++) {
+            const type = types[j];
+            let i = 0;
+            const positions = [];
+            for (let k = 0; k < skips.length; k++) {
+                let delta = +skips[k];
+                do {
+                    if (base === 'N' || base === seq[i]) {
+                        delta--;
+                    }
+                    i++;
+                } while (delta >= 0 && i < seq.length);
+                const temp = i - 1;
+                positions.push(fstrand === -1 ? seq.length - 1 - temp : temp);
+            }
+            if (fstrand === -1) {
+                positions.sort((a, b) => a - b);
+            }
+            result.push({
+                type,
+                positions,
+            });
+        }
+    }
+    return result;
+}
+export function getModificationTypes(mm) {
+    const mods = mm.split(';');
+    return mods
+        .filter(mod => !!mod)
+        .map(mod => {
+        const [basemod] = mod.split(',');
+        const matches = basemod.match(modificationRegex);
+        if (!matches) {
+            throw new Error('bad format for MM tag');
+        }
+        const [, , , typestr] = matches;
+        // can be a multi e.g. C+mh for both meth (m) and hydroxymeth (h) so
+        // split, and they can also be chemical codes (ChEBI) e.g. C+16061
+        return typestr.split(/(\d+|.)/).filter(f => !!f);
+    })
+        .flat();
+}
+//# sourceMappingURL=MismatchParser.js.map