p2s_mapper 1.3.0 → 2.0.0

package/dist/pdbEntry.js

@@ -0,0 +1,182 @@
+"use strict";
+// What RCSB knows about one PDB entry that a mapping needs, in one GraphQL
+// call: each polymer entity's sequence, its author chains, and its alignment
+// to UniProt. The alignment is what SIFTS publishes per chain, numbered the
+// same way (`entity_beg_seq_id` is label_seq_id, 1-based into the SEQRES),
+// and agreed with PDBe's SIFTS on every residue of 23 entries checked
+// (2026-10-11, 1TUP, 2RH1, 4HHB, 7K00 among them), in a tenth of the time.
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.RCSB_DATA_GRAPHQL_URL = void 0;
+exports.pdbEntryQuery = pdbEntryQuery;
+exports.graphqlFetch = graphqlFetch;
+exports.fetchPdbEntry = fetchPdbEntry;
+exports.fetchUniProtStructureMappings = fetchUniProtStructureMappings;
+exports.parsePdbEntry = parsePdbEntry;
+const fetchUtils_ts_1 = require("./fetchUtils.js");
+exports.RCSB_DATA_GRAPHQL_URL = 'https://data.rcsb.org/graphql';
+function pdbEntryQuery(pdbId) {
+    return `{ entry(entry_id: "${pdbId.toUpperCase()}") { polymer_entities {
+  rcsb_id
+  entity_poly { type pdbx_seq_one_letter_code_can }
+  rcsb_polymer_entity_container_identifiers { auth_asym_ids }
+  rcsb_polymer_entity_align {
+    reference_database_name reference_database_accession
+    aligned_regions { entity_beg_seq_id ref_beg_seq_id length }
+  }
+  uniprots { rcsb_id rcsb_uniprot_entry_name }
+} } }`;
+}
+const RETRY_DELAYS_MS = [1000, 3000];
+// bounded, since a view reports itself loading until the entry answers
+const TIMEOUT_MS = 20_000;
+/**
+ * A GraphQL request. The server answers 200 whatever happened, so a failure
+ * is read from the body's `errors`, and counts as a 400: the server
+ * understood the query and declined it, so asking again gets the same answer.
+ */
+async function graphqlFetch(url, query, variables, opts) {
+    const json = await (0, fetchUtils_ts_1.jsonfetch)(url, {
+        ...opts,
+        method: 'POST',
+        headers: { 'Content-Type': 'application/json' },
+        body: JSON.stringify({ query, variables }),
+    });
+    const errors = isRecord(json) ? json.errors : undefined;
+    if (Array.isArray(errors) && errors.length > 0) {
+        const first = errors[0];
+        const message = isRecord(first) && typeof first.message === 'string'
+            ? first.message
+            : 'unknown error';
+        throw new fetchUtils_ts_1.HttpError(`${new URL(url).host}: ${message}`, 400);
+    }
+    return isRecord(json) ? json.data : undefined;
+}
+/**
+ * The entry, retried twice: one failed request would otherwise leave a fusion
+ * construct mapped onto its partner for the whole session. An id RCSB has no
+ * entry for is a 404, which is not retried, since that answer cannot change.
+ */
+async function fetchPdbEntry(pdbId, opts) {
+    const { signal: caller, ...rest } = opts ?? {};
+    const signal = caller ?? undefined;
+    const deadline = () => signal
+        ? AbortSignal.any([signal, AbortSignal.timeout(TIMEOUT_MS)])
+        : AbortSignal.timeout(TIMEOUT_MS);
+    for (let attempt = 0;; attempt++) {
+        try {
+            const entry = parsePdbEntry(await graphqlFetch(exports.RCSB_DATA_GRAPHQL_URL, pdbEntryQuery(pdbId), undefined, { ...rest, signal: deadline() }));
+            if (!entry) {
+                throw new fetchUtils_ts_1.HttpError(`RCSB has no entry ${pdbId.toUpperCase()}`, 404);
+            }
+            return entry;
+        }
+        catch (e) {
+            const delay = RETRY_DELAYS_MS[attempt];
+            // the caller's own abort is not a failure to retry either
+            if (delay === undefined || (0, fetchUtils_ts_1.isDefinitiveFailure)(e) || signal?.aborted) {
+                throw e;
+            }
+            await (0, fetchUtils_ts_1.timeout)(delay, signal);
+        }
+    }
+}
+/** The entry's UniProt alignment alone. */
+async function fetchUniProtStructureMappings(pdbId, opts) {
+    return (await fetchPdbEntry(pdbId, opts)).mappings;
+}
+function isRecord(value) {
+    return typeof value === 'object' && value !== null && !Array.isArray(value);
+}
+function finiteNumber(value) {
+    return typeof value === 'number' && Number.isFinite(value) ? value : undefined;
+}
+function strings(value) {
+    return Array.isArray(value)
+        ? value.filter((v) => typeof v === 'string')
+        : [];
+}
+/**
+ * Parses the `data` of `pdbEntryQuery`. Undefined when RCSB has no such
+ * entry; an entry nothing aligns to UniProt (DNA alone, say) parses to no
+ * mappings. An unparseable alignment is skipped rather than failing the
+ * entry, so a chimera with one malformed region still maps its other chains.
+ */
+function parsePdbEntry(data) {
+    const entry = isRecord(data) ? data.entry : undefined;
+    if (!isRecord(entry)) {
+        return undefined;
+    }
+    const entities = Array.isArray(entry.polymer_entities)
+        ? entry.polymer_entities.filter(isRecord)
+        : [];
+    const byAccession = new Map();
+    const proteinSequences = [];
+    for (const entity of entities.toSorted((a, b) => entityNumber(a.rcsb_id) - entityNumber(b.rcsb_id))) {
+        const entityId = String(entityNumber(entity.rcsb_id));
+        const poly = isRecord(entity.entity_poly) ? entity.entity_poly : undefined;
+        const sequence = poly?.pdbx_seq_one_letter_code_can;
+        if (typeof poly?.type === 'string' &&
+            poly.type.startsWith('polypeptide') &&
+            typeof sequence === 'string' &&
+            sequence.length > 0) {
+            proteinSequences.push(sequence);
+        }
+        const ids = entity.rcsb_polymer_entity_container_identifiers;
+        const chains = strings(isRecord(ids) ? ids.auth_asym_ids : undefined);
+        const names = new Map((Array.isArray(entity.uniprots) ? entity.uniprots : [])
+            .filter(isRecord)
+            .map(u => [u.rcsb_id, strings(u.rcsb_uniprot_entry_name)[0]]));
+        for (const align of Array.isArray(entity.rcsb_polymer_entity_align)
+            ? entity.rcsb_polymer_entity_align.filter(isRecord)
+            : []) {
+            const accession = align.reference_database_accession;
+            if (align.reference_database_name !== 'UniProt' ||
+                typeof accession !== 'string') {
+                continue;
+            }
+            const segments = (Array.isArray(align.aligned_regions) ? align.aligned_regions : []).flatMap(region => parseSegments(region, entityId, chains));
+            if (segments.length === 0) {
+                continue;
+            }
+            const mapping = byAccession.get(accession) ?? {
+                accession,
+                name: names.get(accession),
+                segments: [],
+            };
+            mapping.segments.push(...segments);
+            byAccession.set(accession, mapping);
+        }
+    }
+    return { mappings: [...byAccession.values()], proteinSequences };
+}
+function entityNumber(rcsbId) {
+    return typeof rcsbId === 'string' ? Number(rcsbId.split('_')[1]) : NaN;
+}
+/** One segment per chain carrying the entity, so a PDB-format file, whose
+ * entities Mol* numbers per chain, still matches by chain. */
+function parseSegments(region, entityId, chains) {
+    if (!isRecord(region)) {
+        return [];
+    }
+    const entityBeg = finiteNumber(region.entity_beg_seq_id);
+    const refBeg = finiteNumber(region.ref_beg_seq_id);
+    const length = finiteNumber(region.length);
+    if (entityBeg === undefined ||
+        refBeg === undefined ||
+        length === undefined ||
+        length < 1) {
+        return [];
+    }
+    const segment = {
+        entityId,
+        unpStart: refBeg,
+        unpEnd: refBeg + length - 1,
+        // SEQRES/label_seq_id is 1-based, structure positions are 0-based
+        structStart: entityBeg - 1,
+        structEnd: entityBeg + length - 2,
+    };
+    return chains.length === 0
+        ? [segment]
+        : chains.map(chainId => ({ ...segment, chainId }));
+}
+//# sourceMappingURL=pdbEntry.js.map

package/dist/pdbEntry.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"pdbEntry.js","sourceRoot":"","sources":["../src/pdbEntry.ts"],"names":[],"mappings":";AAAA,2EAA2E;AAC3E,6EAA6E;AAC7E,4EAA4E;AAC5E,2EAA2E;AAC3E,sEAAsE;AACtE,2EAA2E;;;AAyB3E,sCAWC;AAWD,oCAsBC;AAOD,sCAiCC;AAGD,sEAKC;AAsBD,sCAyDC;AAlMD,mDAKwB;AAQX,QAAA,qBAAqB,GAAG,+BAA+B,CAAA;AAUpE,SAAgB,aAAa,CAAC,KAAa;IACzC,OAAO,sBAAsB,KAAK,CAAC,WAAW,EAAE;;;;;;;;;MAS5C,CAAA;AACN,CAAC;AAED,MAAM,eAAe,GAAG,CAAC,IAAI,EAAE,IAAI,CAAC,CAAA;AACpC,uEAAuE;AACvE,MAAM,UAAU,GAAG,MAAM,CAAA;AAEzB;;;;GAIG;AACI,KAAK,UAAU,YAAY,CAChC,GAAW,EACX,KAAa,EACb,SAA8C,EAC9C,IAAmB;IAEnB,MAAM,IAAI,GAAY,MAAM,IAAA,yBAAS,EAAC,GAAG,EAAE;QACzC,GAAG,IAAI;QACP,MAAM,EAAE,MAAM;QACd,OAAO,EAAE,EAAE,cAAc,EAAE,kBAAkB,EAAE;QAC/C,IAAI,EAAE,IAAI,CAAC,SAAS,CAAC,EAAE,KAAK,EAAE,SAAS,EAAE,CAAC;KAC3C,CAAC,CAAA;IACF,MAAM,MAAM,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,SAAS,CAAA;IACvD,IAAI,KAAK,CAAC,OAAO,CAAC,MAAM,CAAC,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC/C,MAAM,KAAK,GAAY,MAAM,CAAC,CAAC,CAAC,CAAA;QAChC,MAAM,OAAO,GACX,QAAQ,CAAC,KAAK,CAAC,IAAI,OAAO,KAAK,CAAC,OAAO,KAAK,QAAQ;YAClD,CAAC,CAAC,KAAK,CAAC,OAAO;YACf,CAAC,CAAC,eAAe,CAAA;QACrB,MAAM,IAAI,yBAAS,CAAC,GAAG,IAAI,GAAG,CAAC,GAAG,CAAC,CAAC,IAAI,KAAK,OAAO,EAAE,EAAE,GAAG,CAAC,CAAA;IAC9D,CAAC;IACD,OAAO,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAC,SAAS,CAAA;AAC/C,CAAC;AAED;;;;GAIG;AACI,KAAK,UAAU,aAAa,CACjC,KAAa,EACb,IAAmB;IAEnB,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,IAAI,EAAE,GAAG,IAAI,IAAI,EAAE,CAAA;IAC9C,MAAM,MAAM,GAAG,MAAM,IAAI,SAAS,CAAA;IAClC,MAAM,QAAQ,GAAG,GAAG,EAAE,CACpB,MAAM;QACJ,CAAC,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,WAAW,CAAC,OAAO,CAAC,UAAU,CAAC,CAAC,CAAC;QAC5D,CAAC,CAAC,WAAW,CAAC,OAAO,CAAC,UAAU,CAAC,CAAA;IACrC,KAAK,IAAI,OAAO,GAAG,CAAC,GAAI,OAAO,EAAE,EAAE,CAAC;QAClC,IAAI,CAAC;YACH,MAAM,KAAK,GAAG,aAAa,CACzB,MAAM,YAAY,CAChB,6BAAqB,EACrB,aAAa,CAAC,KAAK,CAAC,EACpB,SAAS,EACT,EAAE,GAAG,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,EAAE,CAChC,CACF,CAAA;YACD,IAAI,CAAC,KAAK,EAAE,CAAC;gBACX,MAAM,IAAI,yBAAS,CAAC,qBAAqB,KAAK,CAAC,WAAW,EAAE,EAAE,EAAE,GAAG,CAAC,CAAA;YACtE,CAAC;YACD,OAAO,KAAK,CAAA;QACd,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACX,MAAM,KAAK,GAAG,eAAe,CAAC,OAAO,CAAC,CAAA;YACtC,0DAA0D;YAC1D,IAAI,KAAK,KAAK,SAAS,IAAI,IAAA,mCAAmB,EAAC,CAAC,CAAC,IAAI,MAAM,EAAE,OAAO,EAAE,CAAC;gBACrE,MAAM,CAAC,CAAA;YACT,CAAC;YACD,MAAM,IAAA,uBAAO,EAAC,KAAK,EAAE,MAAM,CAAC,CAAA;QAC9B,CAAC;IACH,CAAC;AACH,CAAC;AAED,2CAA2C;AACpC,KAAK,UAAU,6BAA6B,CACjD,KAAa,EACb,IAAmB;IAEnB,OAAO,CAAC,MAAM,aAAa,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC,CAAC,QAAQ,CAAA;AACpD,CAAC;AAED,SAAS,QAAQ,CAAC,KAAc;IAC9B,OAAO,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,CAAA;AAC7E,CAAC;AAED,SAAS,YAAY,CAAC,KAAc;IAClC,OAAO,OAAO,KAAK,KAAK,QAAQ,IAAI,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,SAAS,CAAA;AAChF,CAAC;AAED,SAAS,OAAO,CAAC,KAAc;IAC7B,OAAO,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC;QACzB,CAAC,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAe,EAAE,CAAC,OAAO,CAAC,KAAK,QAAQ,CAAC;QACzD,CAAC,CAAC,EAAE,CAAA;AACR,CAAC;AAED;;;;;GAKG;AACH,SAAgB,aAAa,CAAC,IAAa;IACzC,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,SAAS,CAAA;IACrD,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;QACrB,OAAO,SAAS,CAAA;IAClB,CAAC;IACD,MAAM,QAAQ,GAAG,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,gBAAgB,CAAC;QACpD,CAAC,CAAC,KAAK,CAAC,gBAAgB,CAAC,MAAM,CAAC,QAAQ,CAAC;QACzC,CAAC,CAAC,EAAE,CAAA;IACN,MAAM,WAAW,GAAG,IAAI,GAAG,EAAmC,CAAA;IAC9D,MAAM,gBAAgB,GAAa,EAAE,CAAA;IACrC,KAAK,MAAM,MAAM,IAAI,QAAQ,CAAC,QAAQ,CACpC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,YAAY,CAAC,CAAC,CAAC,OAAO,CAAC,GAAG,YAAY,CAAC,CAAC,CAAC,OAAO,CAAC,CAC5D,EAAE,CAAC;QACF,MAAM,QAAQ,GAAG,MAAM,CAAC,YAAY,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,CAAA;QACrD,MAAM,IAAI,GAAG,QAAQ,CAAC,MAAM,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,WAAW,CAAC,CAAC,CAAC,SAAS,CAAA;QAC1E,MAAM,QAAQ,GAAG,IAAI,EAAE,4BAA4B,CAAA;QACnD,IACE,OAAO,IAAI,EAAE,IAAI,KAAK,QAAQ;YAC9B,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,aAAa,CAAC;YACnC,OAAO,QAAQ,KAAK,QAAQ;YAC5B,QAAQ,CAAC,MAAM,GAAG,CAAC,EACnB,CAAC;YACD,gBAAgB,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;QACjC,CAAC;QACD,MAAM,GAAG,GAAG,MAAM,CAAC,yCAAyC,CAAA;QAC5D,MAAM,MAAM,GAAG,OAAO,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,aAAa,CAAC,CAAC,CAAC,SAAS,CAAC,CAAA;QACrE,MAAM,KAAK,GAAG,IAAI,GAAG,CACnB,CAAC,KAAK,CAAC,OAAO,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,CAAC;aACpD,MAAM,CAAC,QAAQ,CAAC;aAChB,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,OAAO,EAAE,OAAO,CAAC,CAAC,CAAC,uBAAuB,CAAC,CAAC,CAAC,CAAC,CAAU,CAAC,CACzE,CAAA;QACD,KAAK,MAAM,KAAK,IAAI,KAAK,CAAC,OAAO,CAAC,MAAM,CAAC,yBAAyB,CAAC;YACjE,CAAC,CAAC,MAAM,CAAC,yBAAyB,CAAC,MAAM,CAAC,QAAQ,CAAC;YACnD,CAAC,CAAC,EAAE,EAAE,CAAC;YACP,MAAM,SAAS,GAAG,KAAK,CAAC,4BAA4B,CAAA;YACpD,IACE,KAAK,CAAC,uBAAuB,KAAK,SAAS;gBAC3C,OAAO,SAAS,KAAK,QAAQ,EAC7B,CAAC;gBACD,SAAQ;YACV,CAAC;YACD,MAAM,QAAQ,GAAG,CACf,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,eAAe,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,eAAe,CAAC,CAAC,CAAC,EAAE,CAClE,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC,aAAa,CAAC,MAAM,EAAE,QAAQ,EAAE,MAAM,CAAC,CAAC,CAAA;YAC5D,IAAI,QAAQ,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;gBAC1B,SAAQ;YACV,CAAC;YACD,MAAM,OAAO,GAAG,WAAW,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI;gBAC5C,SAAS;gBACT,IAAI,EAAE,KAAK,CAAC,GAAG,CAAC,SAAS,CAAC;gBAC1B,QAAQ,EAAE,EAAE;aACb,CAAA;YACD,OAAO,CAAC,QAAQ,CAAC,IAAI,CAAC,GAAG,QAAQ,CAAC,CAAA;YAClC,WAAW,CAAC,GAAG,CAAC,SAAS,EAAE,OAAO,CAAC,CAAA;QACrC,CAAC;IACH,CAAC;IACD,OAAO,EAAE,QAAQ,EAAE,CAAC,GAAG,WAAW,CAAC,MAAM,EAAE,CAAC,EAAE,gBAAgB,EAAE,CAAA;AAClE,CAAC;AAED,SAAS,YAAY,CAAC,MAAe;IACnC,OAAO,OAAO,MAAM,KAAK,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAA;AACxE,CAAC;AAED;8DAC8D;AAC9D,SAAS,aAAa,CACpB,MAAe,EACf,QAAgB,EAChB,MAAgB;IAEhB,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;QACtB,OAAO,EAAE,CAAA;IACX,CAAC;IACD,MAAM,SAAS,GAAG,YAAY,CAAC,MAAM,CAAC,iBAAiB,CAAC,CAAA;IACxD,MAAM,MAAM,GAAG,YAAY,CAAC,MAAM,CAAC,cAAc,CAAC,CAAA;IAClD,MAAM,MAAM,GAAG,YAAY,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA;IAC1C,IACE,SAAS,KAAK,SAAS;QACvB,MAAM,KAAK,SAAS;QACpB,MAAM,KAAK,SAAS;QACpB,MAAM,GAAG,CAAC,EACV,CAAC;QACD,OAAO,EAAE,CAAA;IACX,CAAC;IACD,MAAM,OAAO,GAAG;QACd,QAAQ;QACR,QAAQ,EAAE,MAAM;QAChB,MAAM,EAAE,MAAM,GAAG,MAAM,GAAG,CAAC;QAC3B,kEAAkE;QAClE,WAAW,EAAE,SAAS,GAAG,CAAC;QAC1B,SAAS,EAAE,SAAS,GAAG,MAAM,GAAG,CAAC;KAClC,CAAA;IACD,OAAO,MAAM,CAAC,MAAM,KAAK,CAAC;QACxB,CAAC,CAAC,CAAC,OAAO,CAAC;QACX,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC,EAAE,GAAG,OAAO,EAAE,OAAO,EAAE,CAAC,CAAC,CAAA;AACtD,CAAC"}

package/dist/pdbUniProtMapping.d.ts

@@ -1,19 +1,14 @@
-import type { FetchOptions } from './fetchUtils.ts';
 /** One contiguous UniProt <-> structure correspondence for a single entity.
  * `unp*` are 1-based UniProt positions; `struct*` are 0-based inclusive
- * structure-sequence positions, this plugin's native coordinate. */
+ * structure-sequence positions, this package's native coordinate. */
 export interface UniProtStructureSegment {
     entityId: string;
-    /** author chain id, when the response names it */
+    /** author chain id, when the source names it */
     chainId?: string;
     unpStart: number;
     unpEnd: number;
     structStart: number;
     structEnd: number;
-    /** the endpoints' author residue numbers, the numbering a paper cites and a
-     * Mol* hover shows; absent when the response omits them */
-    authorStart?: number;
-    authorEnd?: number;
 }
 export interface UniProtStructureMapping {
     accession: string;
@@ -27,27 +22,7 @@ export interface UniProtStructureMapping {
  * the feature instead of drawing it somewhere misleading.
  */
 export type MapUniProtPosition = (uniprotPos: number) => number | undefined;
-export declare function pdbeSiftsUrl(pdbId: string): string;
-/**
- * SIFTS for an entry, retried twice: one failed request would otherwise leave a
- * fusion construct mapped onto its partner for the whole session.
- *
- * A 4xx is not retried. PDBe answers 404 for an entry it has no UniProt
- * mapping for at all — a DNA-only or obsolete entry — and that answer cannot
- * change, so the retries only postponed it: measured 2026-09-16, `1hhb` and
- * `1bna` each took three requests and ~4.8s to say so, which a caller shows as
- * four and a half seconds of "resolving" for a structure that will never
- * resolve.
- */
-export declare function fetchUniProtStructureMappings(pdbId: string, opts?: FetchOptions): Promise<UniProtStructureMapping[]>;
-/**
- * Parses a PDBe `mappings/uniprot/{pdbId}` response into one entry per UniProt
- * accession. Unparseable segments are skipped rather than failing the whole
- * response — a chimera with one malformed segment still maps its other chains.
- * The response is keyed by pdb id, so the single entry is taken whatever its key.
- */
-export declare function parseUniProtStructureMappings(json: unknown): UniProtStructureMapping[];
-/** The entity a SIFTS segment is matched against: its mmCIF id and the author
+/** The entity a segment is matched against: its mmCIF id and the author
  * chains carrying it. */
 export interface SegmentEntity {
     entityId: string;
@@ -63,10 +38,11 @@ export interface SegmentEntity {
  */
 export declare function chooseUniProtMappingForEntity(mappings: UniProtStructureMapping[], entity: SegmentEntity | undefined): UniProtStructureMapping | undefined;
 /**
- * Mapped structure positions that SIFTS assigns to a protein other than the
- * transcript's, on a chain that fuses two. Local alignment bridges a fusion
- * boundary, and on every GPCR fusion construct checked it scattered receptor
- * residues onto the partner, 33 of them onto T4 lysozyme in 2RH1.
+ * Mapped structure positions the entry's UniProt alignment assigns to a
+ * protein other than the transcript's, on a chain that fuses two. Local
+ * alignment bridges a fusion boundary, and on every GPCR fusion construct
+ * checked it scattered receptor residues onto the partner, 33 of them onto T4
+ * lysozyme in 2RH1.
  *
  * `mapped` is each mapped structure position and whether its pair is
  * identical. The transcript's own accession is the one whose covered positions
@@ -75,7 +51,7 @@ export declare function chooseUniProtMappingForEntity(mappings: UniProtStructure
  * fused to CDK2 covers 11 positions against 224 chance hits on the kinase, and
  * counting unmapped the peptide. Unless that accession reaches half identity
  * the transcript is taken to be neither protein, and nothing is unmapped.
- * Residues SIFTS assigns to no accession, such as tags, stay mapped.
+ * Residues assigned to no accession, such as tags, stay mapped.
  */
 export declare function fusionPartnerPositions(mappings: UniProtStructureMapping[], entity: SegmentEntity | undefined, mapped: ReadonlyMap<number, boolean>): Set<number>;
 /** 1-based UniProt position -> 0-based structure position, for an AlphaFold

package/dist/pdbUniProtMapping.js

@@ -1,5 +1,5 @@
 "use strict";
-// SIFTS UniProt <-> PDB residue mapping.
+// UniProt <-> PDB residue mapping.
 //
 // UniProt feature coordinates are positions in the UniProt canonical sequence.
 // For an AlphaFold model that *is* the structure's own sequence, so a feature at
@@ -7,142 +7,24 @@
 // entry it is not: the deposited construct is usually a fragment, often with
 // tags or engineered residues, so the two numberings differ by a per-segment
 // offset (1TUP's p53 chain starts at UniProt 94; 6VXX's spike SEQRES 33 is
-// UniProt 14). SIFTS is the authoritative alignment between the two, published
-// per entity/chain by PDBe.
+// UniProt 14). The entry's own UniProt alignment, which `fetchPdbEntry` reads
+// from RCSB, is the authority on that correspondence.
 //
-// `residue_number` in the PDBe response is the 1-based index into the entity's
-// SEQRES, i.e. label_seq_id, which is exactly this plugin's structure-sequence
-// position + 1 (see applyLociInteractivity.ts for the other place that boundary
-// is crossed).
+// Structure positions here index the entity's SEQRES from 0, i.e.
+// label_seq_id - 1, which is this package's structure-sequence position.
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.identityUniProtPositionMap = void 0;
-exports.pdbeSiftsUrl = pdbeSiftsUrl;
-exports.fetchUniProtStructureMappings = fetchUniProtStructureMappings;
-exports.parseUniProtStructureMappings = parseUniProtStructureMappings;
 exports.chooseUniProtMappingForEntity = chooseUniProtMappingForEntity;
 exports.fusionPartnerPositions = fusionPartnerPositions;
 exports.makeUniProtPositionMap = makeUniProtPositionMap;
-const fetchUtils_ts_1 = require("./fetchUtils.js");
-function pdbeSiftsUrl(pdbId) {
-    return `https://www.ebi.ac.uk/pdbe/api/mappings/uniprot/${pdbId.toLowerCase()}`;
-}
-const SIFTS_RETRY_DELAYS_MS = [1000, 3000];
-// bounded, since a view reports itself loading until SIFTS answers
-const SIFTS_TIMEOUT_MS = 20_000;
-/**
- * SIFTS for an entry, retried twice: one failed request would otherwise leave a
- * fusion construct mapped onto its partner for the whole session.
- *
- * A 4xx is not retried. PDBe answers 404 for an entry it has no UniProt
- * mapping for at all — a DNA-only or obsolete entry — and that answer cannot
- * change, so the retries only postponed it: measured 2026-09-16, `1hhb` and
- * `1bna` each took three requests and ~4.8s to say so, which a caller shows as
- * four and a half seconds of "resolving" for a structure that will never
- * resolve.
- */
-async function fetchUniProtStructureMappings(pdbId, opts) {
-    const { signal: caller, ...rest } = opts ?? {};
-    const signal = caller ?? undefined;
-    const deadline = () => signal
-        ? AbortSignal.any([signal, AbortSignal.timeout(SIFTS_TIMEOUT_MS)])
-        : AbortSignal.timeout(SIFTS_TIMEOUT_MS);
-    for (let attempt = 0;; attempt++) {
-        try {
-            return parseUniProtStructureMappings(await (0, fetchUtils_ts_1.jsonfetch)(pdbeSiftsUrl(pdbId), { ...rest, signal: deadline() }));
-        }
-        catch (e) {
-            const delay = SIFTS_RETRY_DELAYS_MS[attempt];
-            // the caller's own abort is not a failure to retry either
-            if (delay === undefined || (0, fetchUtils_ts_1.isDefinitiveFailure)(e) || signal?.aborted) {
-                throw e;
-            }
-            await (0, fetchUtils_ts_1.timeout)(delay, signal);
-        }
-    }
-}
-function isRecord(value) {
-    return typeof value === 'object' && value !== null && !Array.isArray(value);
-}
-function finiteNumber(value) {
-    return typeof value === 'number' && Number.isFinite(value) ? value : undefined;
-}
-function residueNumber(endpoint) {
-    return isRecord(endpoint) ? finiteNumber(endpoint.residue_number) : undefined;
-}
-function authorResidueNumber(endpoint) {
-    return isRecord(endpoint)
-        ? finiteNumber(endpoint.author_residue_number)
-        : undefined;
-}
-function parseSegment(mapping) {
-    if (!isRecord(mapping)) {
-        return undefined;
-    }
-    const entityId = finiteNumber(mapping.entity_id);
-    const unpStart = finiteNumber(mapping.unp_start);
-    const unpEnd = finiteNumber(mapping.unp_end);
-    const structStart = residueNumber(mapping.start);
-    const structEnd = residueNumber(mapping.end);
-    const authorStart = authorResidueNumber(mapping.start);
-    const authorEnd = authorResidueNumber(mapping.end);
-    return entityId === undefined ||
-        unpStart === undefined ||
-        unpEnd === undefined ||
-        structStart === undefined ||
-        structEnd === undefined
-        ? undefined
-        : {
-            entityId: String(entityId),
-            ...(typeof mapping.chain_id === 'string'
-                ? { chainId: mapping.chain_id }
-                : {}),
-            unpStart,
-            unpEnd,
-            // SEQRES/label_seq_id is 1-based, structure positions are 0-based
-            structStart: structStart - 1,
-            structEnd: structEnd - 1,
-            ...(authorStart === undefined ? {} : { authorStart }),
-            ...(authorEnd === undefined ? {} : { authorEnd }),
-        };
-}
-/**
- * Parses a PDBe `mappings/uniprot/{pdbId}` response into one entry per UniProt
- * accession. Unparseable segments are skipped rather than failing the whole
- * response — a chimera with one malformed segment still maps its other chains.
- * The response is keyed by pdb id, so the single entry is taken whatever its key.
- */
-function parseUniProtStructureMappings(json) {
-    const entry = isRecord(json) ? Object.values(json)[0] : undefined;
-    const uniprot = isRecord(entry) ? entry.UniProt : undefined;
-    if (!isRecord(uniprot)) {
-        return [];
-    }
-    const result = [];
-    for (const [accession, value] of Object.entries(uniprot)) {
-        const mappings = isRecord(value) ? value.mappings : undefined;
-        const segments = Array.isArray(mappings)
-            ? mappings.map(parseSegment).filter(s => s !== undefined)
-            : [];
-        if (segments.length > 0) {
-            result.push({
-                accession,
-                name: isRecord(value) && typeof value.name === 'string'
-                    ? value.name
-                    : undefined,
-                segments,
-            });
-        }
-    }
-    return result;
-}
 function coveredResidues(segments) {
     return segments.reduce((a, s) => a + (s.structEnd - s.structStart + 1), 0);
 }
 /**
- * The segments of one accession that describe this entity. SIFTS numbers
- * entities as the mmCIF does, and a PDB-format file loaded through Mol* numbers
- * them per chain (1H26.pdb's chain C is entity 2, which in SIFTS is cyclin A),
- * so a segment naming its chain is matched by chain.
+ * The segments of one accession that describe this entity. The mapping
+ * numbers entities as the mmCIF does, and a PDB-format file loaded through
+ * Mol* numbers them per chain (1H26.pdb's chain C is entity 2, which in the
+ * mmCIF is cyclin A), so a segment naming its chain is matched by chain.
  */
 function segmentsForEntity(mapping, entity) {
     return dedupeSegments(mapping.segments.filter(s => s.chainId === undefined
@@ -174,10 +56,11 @@ function chooseUniProtMappingForEntity(mappings, entity) {
 const OWN_PSEUDOCOUNT = 5;
 const MIN_OWN_IDENTITY = 0.5;
 /**
- * Mapped structure positions that SIFTS assigns to a protein other than the
- * transcript's, on a chain that fuses two. Local alignment bridges a fusion
- * boundary, and on every GPCR fusion construct checked it scattered receptor
- * residues onto the partner, 33 of them onto T4 lysozyme in 2RH1.
+ * Mapped structure positions the entry's UniProt alignment assigns to a
+ * protein other than the transcript's, on a chain that fuses two. Local
+ * alignment bridges a fusion boundary, and on every GPCR fusion construct
+ * checked it scattered receptor residues onto the partner, 33 of them onto T4
+ * lysozyme in 2RH1.
  *
  * `mapped` is each mapped structure position and whether its pair is
  * identical. The transcript's own accession is the one whose covered positions
@@ -186,7 +69,7 @@ const MIN_OWN_IDENTITY = 0.5;
  * fused to CDK2 covers 11 positions against 224 chance hits on the kinase, and
  * counting unmapped the peptide. Unless that accession reaches half identity
  * the transcript is taken to be neither protein, and nothing is unmapped.
- * Residues SIFTS assigns to no accession, such as tags, stay mapped.
+ * Residues assigned to no accession, such as tags, stay mapped.
  */
 function fusionPartnerPositions(mappings, entity, mapped) {
     const partners = new Set();
@@ -223,8 +106,8 @@ function fusionPartnerPositions(mappings, entity, mapped) {
     }
     return partners;
 }
-// SIFTS lists one mapping per *chain*, so the several chains of a homodimer
-// repeat the same entity-level correspondence. They collapse to one segment.
+// The mapping lists one segment per *chain*, so the several chains of a
+// homodimer repeat the same entity-level correspondence. They collapse to one.
 function dedupeSegments(segments) {
     const seen = new Set();
     return segments.filter(s => {

package/dist/pdbUniProtMapping.js.map

@@ -1 +1 @@
-{"version":3,"file":"pdbUniProtMapping.js","sourceRoot":"","sources":["../src/pdbUniProtMapping.ts"],"names":[],"mappings":";AAAA,yCAAyC;AACzC,EAAE;AACF,+EAA+E;AAC/E,iFAAiF;AACjF,+EAA+E;AAC/E,6EAA6E;AAC7E,6EAA6E;AAC7E,2EAA2E;AAC3E,+EAA+E;AAC/E,4BAA4B;AAC5B,EAAE;AACF,+EAA+E;AAC/E,+EAA+E;AAC/E,gFAAgF;AAChF,eAAe;;;AAqCf,oCAEC;AAiBD,sEAwBC;AA0DD,sEA0BC;AAwCD,sEAkBC;AAoBD,wDAyCC;AAmBD,wDAaC;AAzTD,mDAAyE;AAmCzE,SAAgB,YAAY,CAAC,KAAa;IACxC,OAAO,mDAAmD,KAAK,CAAC,WAAW,EAAE,EAAE,CAAA;AACjF,CAAC;AAED,MAAM,qBAAqB,GAAG,CAAC,IAAI,EAAE,IAAI,CAAC,CAAA;AAC1C,mEAAmE;AACnE,MAAM,gBAAgB,GAAG,MAAM,CAAA;AAE/B;;;;;;;;;;GAUG;AACI,KAAK,UAAU,6BAA6B,CACjD,KAAa,EACb,IAAmB;IAEnB,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,IAAI,EAAE,GAAG,IAAI,IAAI,EAAE,CAAA;IAC9C,MAAM,MAAM,GAAG,MAAM,IAAI,SAAS,CAAA;IAClC,MAAM,QAAQ,GAAG,GAAG,EAAE,CACpB,MAAM;QACJ,CAAC,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,WAAW,CAAC,OAAO,CAAC,gBAAgB,CAAC,CAAC,CAAC;QAClE,CAAC,CAAC,WAAW,CAAC,OAAO,CAAC,gBAAgB,CAAC,CAAA;IAC3C,KAAK,IAAI,OAAO,GAAG,CAAC,GAAI,OAAO,EAAE,EAAE,CAAC;QAClC,IAAI,CAAC;YACH,OAAO,6BAA6B,CAClC,MAAM,IAAA,yBAAS,EAAC,YAAY,CAAC,KAAK,CAAC,EAAE,EAAE,GAAG,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,EAAE,CAAC,CACtE,CAAA;QACH,CAAC;QAAC,OAAO,CAAC,EAAE,CAAC;YACX,MAAM,KAAK,GAAG,qBAAqB,CAAC,OAAO,CAAC,CAAA;YAC5C,0DAA0D;YAC1D,IAAI,KAAK,KAAK,SAAS,IAAI,IAAA,mCAAmB,EAAC,CAAC,CAAC,IAAI,MAAM,EAAE,OAAO,EAAE,CAAC;gBACrE,MAAM,CAAC,CAAA;YACT,CAAC;YACD,MAAM,IAAA,uBAAO,EAAC,KAAK,EAAE,MAAM,CAAC,CAAA;QAC9B,CAAC;IACH,CAAC;AACH,CAAC;AAED,SAAS,QAAQ,CAAC,KAAc;IAC9B,OAAO,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,CAAA;AAC7E,CAAC;AAED,SAAS,YAAY,CAAC,KAAc;IAClC,OAAO,OAAO,KAAK,KAAK,QAAQ,IAAI,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,SAAS,CAAA;AAChF,CAAC;AAED,SAAS,aAAa,CAAC,QAAiB;IACtC,OAAO,QAAQ,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,YAAY,CAAC,QAAQ,CAAC,cAAc,CAAC,CAAC,CAAC,CAAC,SAAS,CAAA;AAC/E,CAAC;AAED,SAAS,mBAAmB,CAAC,QAAiB;IAC5C,OAAO,QAAQ,CAAC,QAAQ,CAAC;QACvB,CAAC,CAAC,YAAY,CAAC,QAAQ,CAAC,qBAAqB,CAAC;QAC9C,CAAC,CAAC,SAAS,CAAA;AACf,CAAC;AAED,SAAS,YAAY,CAAC,OAAgB;IACpC,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACvB,OAAO,SAAS,CAAA;IAClB,CAAC;IACD,MAAM,QAAQ,GAAG,YAAY,CAAC,OAAO,CAAC,SAAS,CAAC,CAAA;IAChD,MAAM,QAAQ,GAAG,YAAY,CAAC,OAAO,CAAC,SAAS,CAAC,CAAA;IAChD,MAAM,MAAM,GAAG,YAAY,CAAC,OAAO,CAAC,OAAO,CAAC,CAAA;IAC5C,MAAM,WAAW,GAAG,aAAa,CAAC,OAAO,CAAC,KAAK,CAAC,CAAA;IAChD,MAAM,SAAS,GAAG,aAAa,CAAC,OAAO,CAAC,GAAG,CAAC,CAAA;IAC5C,MAAM,WAAW,GAAG,mBAAmB,CAAC,OAAO,CAAC,KAAK,CAAC,CAAA;IACtD,MAAM,SAAS,GAAG,mBAAmB,CAAC,OAAO,CAAC,GAAG,CAAC,CAAA;IAClD,OAAO,QAAQ,KAAK,SAAS;QAC3B,QAAQ,KAAK,SAAS;QACtB,MAAM,KAAK,SAAS;QACpB,WAAW,KAAK,SAAS;QACzB,SAAS,KAAK,SAAS;QACvB,CAAC,CAAC,SAAS;QACX,CAAC,CAAC;YACE,QAAQ,EAAE,MAAM,CAAC,QAAQ,CAAC;YAC1B,GAAG,CAAC,OAAO,OAAO,CAAC,QAAQ,KAAK,QAAQ;gBACtC,CAAC,CAAC,EAAE,OAAO,EAAE,OAAO,CAAC,QAAQ,EAAE;gBAC/B,CAAC,CAAC,EAAE,CAAC;YACP,QAAQ;YACR,MAAM;YACN,kEAAkE;YAClE,WAAW,EAAE,WAAW,GAAG,CAAC;YAC5B,SAAS,EAAE,SAAS,GAAG,CAAC;YACxB,GAAG,CAAC,WAAW,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,WAAW,EAAE,CAAC;YACrD,GAAG,CAAC,SAAS,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,SAAS,EAAE,CAAC;SAClD,CAAA;AACP,CAAC;AAED;;;;;GAKG;AACH,SAAgB,6BAA6B,CAC3C,IAAa;IAEb,MAAM,KAAK,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAA;IACjE,MAAM,OAAO,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,SAAS,CAAA;IAC3D,IAAI,CAAC,QAAQ,CAAC,OAAO,CAAC,EAAE,CAAC;QACvB,OAAO,EAAE,CAAA;IACX,CAAC;IACD,MAAM,MAAM,GAA8B,EAAE,CAAA;IAC5C,KAAK,MAAM,CAAC,SAAS,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,OAAO,CAAC,EAAE,CAAC;QACzD,MAAM,QAAQ,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,CAAC,SAAS,CAAA;QAC7D,MAAM,QAAQ,GAAG,KAAK,CAAC,OAAO,CAAC,QAAQ,CAAC;YACtC,CAAC,CAAC,QAAQ,CAAC,GAAG,CAAC,YAAY,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,KAAK,SAAS,CAAC;YACzD,CAAC,CAAC,EAAE,CAAA;QACN,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACxB,MAAM,CAAC,IAAI,CAAC;gBACV,SAAS;gBACT,IAAI,EACF,QAAQ,CAAC,KAAK,CAAC,IAAI,OAAO,KAAK,CAAC,IAAI,KAAK,QAAQ;oBAC/C,CAAC,CAAC,KAAK,CAAC,IAAI;oBACZ,CAAC,CAAC,SAAS;gBACf,QAAQ;aACT,CAAC,CAAA;QACJ,CAAC;IACH,CAAC;IACD,OAAO,MAAM,CAAA;AACf,CAAC;AAED,SAAS,eAAe,CAAC,QAAmC;IAC1D,OAAO,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,SAAS,GAAG,CAAC,CAAC,WAAW,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,CAAA;AAC5E,CAAC;AASD;;;;;GAKG;AACH,SAAS,iBAAiB,CACxB,OAAgC,EAChC,MAAqB;IAErB,OAAO,cAAc,CACnB,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAC1B,CAAC,CAAC,OAAO,KAAK,SAAS;QACrB,CAAC,CAAC,CAAC,CAAC,QAAQ,KAAK,MAAM,CAAC,QAAQ;QAChC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,CACtC,CACF,CAAA;AACH,CAAC;AAED;;;;;;;GAOG;AACH,SAAgB,6BAA6B,CAC3C,QAAmC,EACnC,MAAiC;IAEjC,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;QACzB,OAAO,SAAS,CAAA;IAClB,CAAC;IACD,IAAI,IAAyC,CAAA;IAC7C,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;QAC/B,MAAM,QAAQ,GAAG,iBAAiB,CAAC,OAAO,EAAE,MAAM,CAAC,CAAA;QACnD,IACE,QAAQ,CAAC,MAAM,GAAG,CAAC;YACnB,CAAC,CAAC,IAAI,IAAI,eAAe,CAAC,QAAQ,CAAC,GAAG,eAAe,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,EACrE,CAAC;YACD,IAAI,GAAG,EAAE,GAAG,OAAO,EAAE,QAAQ,EAAE,CAAA;QACjC,CAAC;IACH,CAAC;IACD,OAAO,IAAI,CAAA;AACb,CAAC;AAED,MAAM,eAAe,GAAG,CAAC,CAAA;AACzB,MAAM,gBAAgB,GAAG,GAAG,CAAA;AAE5B;;;;;;;;;;;;;;GAcG;AACH,SAAgB,sBAAsB,CACpC,QAAmC,EACnC,MAAiC,EACjC,MAAoC;IAEpC,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAU,CAAA;IAClC,MAAM,WAAW,GAAG,MAAM;QACxB,CAAC,CAAC,QAAQ;aACL,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,iBAAiB,CAAC,CAAC,EAAE,MAAM,CAAC,CAAC;aACtC,MAAM,CAAC,QAAQ,CAAC,EAAE,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC;QAC5C,CAAC,CAAC,EAAE,CAAA;IACN,IAAI,WAAW,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC3B,OAAO,QAAQ,CAAA;IACjB,CAAC;IACD,MAAM,MAAM,GAAG,CAAC,QAAmC,EAAE,GAAW,EAAE,EAAE,CAClE,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,CAAC,WAAW,IAAI,GAAG,IAAI,CAAC,CAAC,SAAS,CAAC,CAAA;IAChE,MAAM,QAAQ,GAAG,WAAW,CAAC,GAAG,CAAC,QAAQ,CAAC,EAAE;QAC1C,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,SAAS,GAAG,CAAC,CAAA;QACjB,KAAK,MAAM,CAAC,GAAG,EAAE,IAAI,CAAC,IAAI,MAAM,EAAE,CAAC;YACjC,IAAI,MAAM,CAAC,QAAQ,EAAE,GAAG,CAAC,EAAE,CAAC;gBAC1B,OAAO,EAAE,CAAA;gBACT,SAAS,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;YAC3B,CAAC;QACH,CAAC;QACD,OAAO,SAAS,GAAG,CAAC,OAAO,GAAG,eAAe,CAAC,CAAA;IAChD,CAAC,CAAC,CAAA;IACF,MAAM,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,QAAQ,CAAC,CAAA;IACzC,IAAI,WAAW,GAAG,gBAAgB,EAAE,CAAC;QACnC,OAAO,QAAQ,CAAA;IACjB,CAAC;IACD,MAAM,GAAG,GAAG,WAAW,CAAC,QAAQ,CAAC,OAAO,CAAC,WAAW,CAAC,CAAE,CAAA;IACvD,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,IAAI,EAAE,EAAE,CAAC;QAChC,IACE,CAAC,MAAM,CAAC,GAAG,EAAE,GAAG,CAAC;YACjB,WAAW,CAAC,IAAI,CAAC,QAAQ,CAAC,EAAE,CAAC,QAAQ,KAAK,GAAG,IAAI,MAAM,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC,EACvE,CAAC;YACD,QAAQ,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;QACnB,CAAC;IACH,CAAC;IACD,OAAO,QAAQ,CAAA;AACjB,CAAC;AAED,4EAA4E;AAC5E,6EAA6E;AAC7E,SAAS,cAAc,CAAC,QAAmC;IACzD,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAA;IAC9B,OAAO,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE;QACzB,MAAM,GAAG,GAAG,GAAG,CAAC,CAAC,QAAQ,IAAI,CAAC,CAAC,MAAM,IAAI,CAAC,CAAC,WAAW,IAAI,CAAC,CAAC,SAAS,EAAE,CAAA;QACvE,MAAM,KAAK,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;QAC5B,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;QACb,OAAO,KAAK,CAAA;IACd,CAAC,CAAC,CAAA;AACJ,CAAC;AAED;mDACmD;AAC5C,MAAM,0BAA0B,GAAuB,UAAU,CAAC,EAAE,CACzE,UAAU,GAAG,CAAC,CAAA;AADH,QAAA,0BAA0B,8BACvB;AAEhB,SAAgB,sBAAsB,CACpC,QAAmC;IAEnC,OAAO,UAAU,CAAC,EAAE;QAClB,MAAM,OAAO,GAAG,QAAQ,CAAC,IAAI,CAC3B,CAAC,CAAC,EAAE,CAAC,UAAU,IAAI,CAAC,CAAC,QAAQ,IAAI,UAAU,IAAI,CAAC,CAAC,MAAM,CACxD,CAAA;QACD,IAAI,CAAC,OAAO,EAAE,CAAC;YACb,OAAO,SAAS,CAAA;QAClB,CAAC;QACD,MAAM,GAAG,GAAG,OAAO,CAAC,WAAW,GAAG,CAAC,UAAU,GAAG,OAAO,CAAC,QAAQ,CAAC,CAAA;QACjE,OAAO,GAAG,IAAI,OAAO,CAAC,SAAS,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,SAAS,CAAA;IACnD,CAAC,CAAA;AACH,CAAC"}
+{"version":3,"file":"pdbUniProtMapping.js","sourceRoot":"","sources":["../src/pdbUniProtMapping.ts"],"names":[],"mappings":";AAAA,mCAAmC;AACnC,EAAE;AACF,+EAA+E;AAC/E,iFAAiF;AACjF,+EAA+E;AAC/E,6EAA6E;AAC7E,6EAA6E;AAC7E,2EAA2E;AAC3E,8EAA8E;AAC9E,sDAAsD;AACtD,EAAE;AACF,kEAAkE;AAClE,yEAAyE;;;AAmEzE,sEAkBC;AAqBD,wDAyCC;AAmBD,wDAaC;AAtJD,SAAS,eAAe,CAAC,QAAmC;IAC1D,OAAO,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,SAAS,GAAG,CAAC,CAAC,WAAW,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,CAAA;AAC5E,CAAC;AASD;;;;;GAKG;AACH,SAAS,iBAAiB,CACxB,OAAgC,EAChC,MAAqB;IAErB,OAAO,cAAc,CACnB,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAC1B,CAAC,CAAC,OAAO,KAAK,SAAS;QACrB,CAAC,CAAC,CAAC,CAAC,QAAQ,KAAK,MAAM,CAAC,QAAQ;QAChC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC,CACtC,CACF,CAAA;AACH,CAAC;AAED;;;;;;;GAOG;AACH,SAAgB,6BAA6B,CAC3C,QAAmC,EACnC,MAAiC;IAEjC,IAAI,MAAM,KAAK,SAAS,EAAE,CAAC;QACzB,OAAO,SAAS,CAAA;IAClB,CAAC;IACD,IAAI,IAAyC,CAAA;IAC7C,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;QAC/B,MAAM,QAAQ,GAAG,iBAAiB,CAAC,OAAO,EAAE,MAAM,CAAC,CAAA;QACnD,IACE,QAAQ,CAAC,MAAM,GAAG,CAAC;YACnB,CAAC,CAAC,IAAI,IAAI,eAAe,CAAC,QAAQ,CAAC,GAAG,eAAe,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC,EACrE,CAAC;YACD,IAAI,GAAG,EAAE,GAAG,OAAO,EAAE,QAAQ,EAAE,CAAA;QACjC,CAAC;IACH,CAAC;IACD,OAAO,IAAI,CAAA;AACb,CAAC;AAED,MAAM,eAAe,GAAG,CAAC,CAAA;AACzB,MAAM,gBAAgB,GAAG,GAAG,CAAA;AAE5B;;;;;;;;;;;;;;;GAeG;AACH,SAAgB,sBAAsB,CACpC,QAAmC,EACnC,MAAiC,EACjC,MAAoC;IAEpC,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAU,CAAA;IAClC,MAAM,WAAW,GAAG,MAAM;QACxB,CAAC,CAAC,QAAQ;aACL,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,iBAAiB,CAAC,CAAC,EAAE,MAAM,CAAC,CAAC;aACtC,MAAM,CAAC,QAAQ,CAAC,EAAE,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC;QAC5C,CAAC,CAAC,EAAE,CAAA;IACN,IAAI,WAAW,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC3B,OAAO,QAAQ,CAAA;IACjB,CAAC;IACD,MAAM,MAAM,GAAG,CAAC,QAAmC,EAAE,GAAW,EAAE,EAAE,CAClE,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,CAAC,WAAW,IAAI,GAAG,IAAI,CAAC,CAAC,SAAS,CAAC,CAAA;IAChE,MAAM,QAAQ,GAAG,WAAW,CAAC,GAAG,CAAC,QAAQ,CAAC,EAAE;QAC1C,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,SAAS,GAAG,CAAC,CAAA;QACjB,KAAK,MAAM,CAAC,GAAG,EAAE,IAAI,CAAC,IAAI,MAAM,EAAE,CAAC;YACjC,IAAI,MAAM,CAAC,QAAQ,EAAE,GAAG,CAAC,EAAE,CAAC;gBAC1B,OAAO,EAAE,CAAA;gBACT,SAAS,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;YAC3B,CAAC;QACH,CAAC;QACD,OAAO,SAAS,GAAG,CAAC,OAAO,GAAG,eAAe,CAAC,CAAA;IAChD,CAAC,CAAC,CAAA;IACF,MAAM,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,QAAQ,CAAC,CAAA;IACzC,IAAI,WAAW,GAAG,gBAAgB,EAAE,CAAC;QACnC,OAAO,QAAQ,CAAA;IACjB,CAAC;IACD,MAAM,GAAG,GAAG,WAAW,CAAC,QAAQ,CAAC,OAAO,CAAC,WAAW,CAAC,CAAE,CAAA;IACvD,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,IAAI,EAAE,EAAE,CAAC;QAChC,IACE,CAAC,MAAM,CAAC,GAAG,EAAE,GAAG,CAAC;YACjB,WAAW,CAAC,IAAI,CAAC,QAAQ,CAAC,EAAE,CAAC,QAAQ,KAAK,GAAG,IAAI,MAAM,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC,EACvE,CAAC;YACD,QAAQ,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;QACnB,CAAC;IACH,CAAC;IACD,OAAO,QAAQ,CAAA;AACjB,CAAC;AAED,wEAAwE;AACxE,+EAA+E;AAC/E,SAAS,cAAc,CAAC,QAAmC;IACzD,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAA;IAC9B,OAAO,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE;QACzB,MAAM,GAAG,GAAG,GAAG,CAAC,CAAC,QAAQ,IAAI,CAAC,CAAC,MAAM,IAAI,CAAC,CAAC,WAAW,IAAI,CAAC,CAAC,SAAS,EAAE,CAAA;QACvE,MAAM,KAAK,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;QAC5B,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;QACb,OAAO,KAAK,CAAA;IACd,CAAC,CAAC,CAAA;AACJ,CAAC;AAED;mDACmD;AAC5C,MAAM,0BAA0B,GAAuB,UAAU,CAAC,EAAE,CACzE,UAAU,GAAG,CAAC,CAAA;AADH,QAAA,0BAA0B,8BACvB;AAEhB,SAAgB,sBAAsB,CACpC,QAAmC;IAEnC,OAAO,UAAU,CAAC,EAAE;QAClB,MAAM,OAAO,GAAG,QAAQ,CAAC,IAAI,CAC3B,CAAC,CAAC,EAAE,CAAC,UAAU,IAAI,CAAC,CAAC,QAAQ,IAAI,UAAU,IAAI,CAAC,CAAC,MAAM,CACxD,CAAA;QACD,IAAI,CAAC,OAAO,EAAE,CAAC;YACb,OAAO,SAAS,CAAA;QAClB,CAAC;QACD,MAAM,GAAG,GAAG,OAAO,CAAC,WAAW,GAAG,CAAC,UAAU,GAAG,OAAO,CAAC,QAAQ,CAAC,CAAA;QACjE,OAAO,GAAG,IAAI,OAAO,CAAC,SAAS,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,SAAS,CAAA;IACnD,CAAC,CAAA;AACH,CAAC"}

package/dist/structureSources.d.ts

@@ -1,6 +1,8 @@
 import type { FetchOptions } from './fetchUtils.ts';
+export declare const RCSB_SEQUENCE_COORDINATES_URL = "https://sequence-coordinates.rcsb.org/graphql";
 export interface ExperimentalStructure {
     pdbId: string;
+    /** RCSB's name for it: 'X-ray', 'EM', 'NMR', ... */
     method: string;
     /** Å; absent for NMR and some EM entries */
     resolution?: number;
@@ -9,20 +11,42 @@ export interface ExperimentalStructure {
     end: number;
     /** fraction of the UniProt sequence the entry covers, 0..1 */
     coverage: number;
+    /** author chain ids carrying the protein */
+    chains: string[];
 }
-export declare function beaconsSummaryUrl(uniprotId: string): string;
+/** One polymer entity aligned to the accession. */
+export interface EntityAlignment {
+    /** RCSB's id, "1TUP_3" */
+    entityId: string;
+    pdbId: string;
+    start: number;
+    end: number;
+    coverage: number;
+}
+export interface EntityDetails {
+    method: string;
+    resolution?: number;
+    chains: string[];
+}
+export declare function entityAlignmentsQuery(uniprotId: string): string;
+export declare const ENTITY_DETAILS_QUERY = "query($ids: [String!]!) { polymer_entities(entity_ids: $ids) {\n  rcsb_id\n  rcsb_polymer_entity_container_identifiers { auth_asym_ids }\n  entry { rcsb_entry_info { experimental_method resolution_combined } }\n} }";
+/**
+ * PDB entities only: the same list carries RCSB's computed models
+ * (`AF_AFP04637F1_1`), which a residue mapping does not exist for. The range
+ * spans every aligned region, as 3D-Beacons reported it.
+ */
+export declare function parseEntityAlignments(data: unknown): EntityAlignment[];
+export declare function parseEntityDetails(data: unknown): Map<string, EntityDetails>;
+/**
+ * One row per entry, best coverage first and sharpest within a tie. An entry
+ * with several entities of the protein (a construct crystallised beside a
+ * different fragment of itself) takes the best-covering entity's range and
+ * every entity's chains.
+ */
+export declare function experimentalStructures(alignments: EntityAlignment[], details: ReadonlyMap<string, EntityDetails>): ExperimentalStructure[];
 /**
- * PDBe entries only, best-covering first and sharpest within a tie.
- *
- * Predicted entries (AlphaFold, SWISS-MODEL, AlphaFill) are left out because a
- * prediction comes from AlphaFold's own API, which answers per model. And
- * 3D-Beacons files SASBDB's small-angle-scattering fits under "experimentally
- * determined" too, with numeric ids and near-full coverage — dystrophin's best
- * "structure" by coverage was SASBDB 436 — which are not entries a residue
- * mapping exists for. Hence the `provider === 'PDBe'` gate rather than the
- * category alone.
+ * Every PDB entry aligned to the accession. An accession with none, or one
+ * RCSB does not know, is `[]`; an isoform accession is asked as its entry,
+ * which is what the alignment is numbered on.
  */
-export declare function parseExperimentalStructures(json: unknown): ExperimentalStructure[];
-/** Best-effort: an unreachable API reads as "no entries", which costs a
- * structure suggestion and nothing else. */
 export declare function fetchExperimentalStructures(uniprotId: string, opts?: FetchOptions): Promise<ExperimentalStructure[]>;