p2s_mapper 1.3.0 → 2.0.0

package/src/structureSources.ts

@@ -1,16 +1,21 @@
-// The experimental structures 3D-Beacons knows for a UniProt entry: every PDBe
-// entry mapped to the accession, with the UniProt range it covers and its
-// resolution, in one CORS-enabled call. That is what lets a page offer a
-// crystal structure beside a prediction without a PDB search of its own; take
-// the entry id from here and ask `fetchUniProtStructureMappings` for its
+// The experimental structures RCSB aligns to a UniProt entry: every PDB entity
+// carrying the accession, with the UniProt range it covers, then each entry's
+// method, resolution and chains. Two CORS-enabled calls, which is what lets a
+// page offer a crystal structure beside a prediction without a PDB search of
+// its own; take the entry id from here and ask `fetchPdbEntry` for its
 // residue mapping.
 
-import { rawfetch } from './fetchUtils.ts'
+import { RCSB_DATA_GRAPHQL_URL, graphqlFetch } from './pdbEntry.ts'
+import { isPdbId } from './structureUrls.ts'
 
 import type { FetchOptions } from './fetchUtils.ts'
 
+export 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
@@ -19,15 +24,45 @@ 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[]
 }
 
-// Unfiltered: the `?provider=pdbe` form answered BRAF, HBB and BRCA2 with a 404
-// about one time in three (2026-10-09), and the parser keeps PDBe's entries
-// anyway.
-export function beaconsSummaryUrl(uniprotId: string) {
-  return `https://www.ebi.ac.uk/pdbe/pdbe-kb/3dbeacons/api/uniprot/summary/${encodeURIComponent(uniprotId)}.json`
+/** 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 function entityAlignmentsQuery(uniprotId: string) {
+  return `{ alignments(from: UNIPROT, to: PDB_ENTITY, queryId: "${uniprotId}") {
+  target_alignments {
+    target_id
+    coverage { query_coverage }
+    aligned_regions { query_begin query_end }
+  }
+} }`
+}
+
+export const ENTITY_DETAILS_QUERY = `query($ids: [String!]!) { polymer_entities(entity_ids: $ids) {
+  rcsb_id
+  rcsb_polymer_entity_container_identifiers { auth_asym_ids }
+  entry { rcsb_entry_info { experimental_method resolution_combined } }
+} }`
+
+// RCSB's cap on ids per batch request
+const DETAILS_BATCH = 1000
+
 function isRecord(value: unknown): value is Record<string, unknown> {
   return typeof value === 'object' && value !== null && !Array.isArray(value)
 }
@@ -36,77 +71,166 @@ function finiteNumber(value: unknown) {
   return typeof value === 'number' && Number.isFinite(value) ? value : undefined
 }
 
+function strings(value: unknown) {
+  return Array.isArray(value)
+    ? value.filter((v): v is string => typeof v === 'string')
+    : []
+}
+
 /**
- * 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.
+ * 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 function parseExperimentalStructures(
-  json: unknown,
-): ExperimentalStructure[] {
-  const structures = isRecord(json) ? json.structures : undefined
-  if (!Array.isArray(structures)) {
+export function parseEntityAlignments(data: unknown): EntityAlignment[] {
+  const alignments = isRecord(data) ? data.alignments : undefined
+  const targets = isRecord(alignments)
+    ? alignments.target_alignments
+    : undefined
+  if (!Array.isArray(targets)) {
     return []
   }
-  return structures
-    .flatMap(entry => {
-      const s = isRecord(entry) ? entry.summary : undefined
-      if (!isRecord(s)) {
-        return []
-      }
-      const start = finiteNumber(s.uniprot_start)
-      const end = finiteNumber(s.uniprot_end)
-      const coverage = finiteNumber(s.coverage)
-      return typeof s.model_identifier === 'string' &&
-        s.provider === 'PDBe' &&
-        s.model_category === 'EXPERIMENTALLY DETERMINED' &&
-        start !== undefined &&
-        end !== undefined &&
-        coverage !== undefined
-        ? [
-            {
-              pdbId: s.model_identifier.toLowerCase(),
-              method:
-                typeof s.experimental_method === 'string'
-                  ? s.experimental_method
-                  : 'experimental',
-              resolution: finiteNumber(s.resolution),
-              start,
-              end,
-              coverage,
-            },
-          ]
-        : []
+  return targets.flatMap(target => {
+    if (!isRecord(target) || typeof target.target_id !== 'string') {
+      return []
+    }
+    const [pdbId] = target.target_id.split('_')
+    const coverage = isRecord(target.coverage)
+      ? finiteNumber(target.coverage.query_coverage)
+      : undefined
+    const regions = Array.isArray(target.aligned_regions)
+      ? target.aligned_regions.filter(isRecord)
+      : []
+    const begins = regions.map(r => finiteNumber(r.query_begin))
+    const ends = regions.map(r => finiteNumber(r.query_end))
+    return pdbId !== undefined &&
+      isPdbId(pdbId) &&
+      coverage !== undefined &&
+      begins.length > 0 &&
+      begins.every(b => b !== undefined) &&
+      ends.every(e => e !== undefined)
+      ? [
+          {
+            entityId: target.target_id,
+            pdbId: pdbId.toLowerCase(),
+            start: Math.min(...begins),
+            end: Math.max(...ends),
+            coverage,
+          },
+        ]
+      : []
+  })
+}
+
+export function parseEntityDetails(data: unknown): Map<string, EntityDetails> {
+  const entities = isRecord(data) ? data.polymer_entities : undefined
+  const details = new Map<string, EntityDetails>()
+  for (const entity of Array.isArray(entities) ? entities : []) {
+    if (!isRecord(entity) || typeof entity.rcsb_id !== 'string') {
+      continue
+    }
+    const ids = entity.rcsb_polymer_entity_container_identifiers
+    const info = isRecord(entity.entry)
+      ? entity.entry.rcsb_entry_info
+      : undefined
+    const resolutions = isRecord(info) ? info.resolution_combined : undefined
+    const resolution = Array.isArray(resolutions)
+      ? finiteNumber(resolutions[0])
+      : undefined
+    details.set(entity.rcsb_id, {
+      method:
+        isRecord(info) && typeof info.experimental_method === 'string'
+          ? info.experimental_method
+          : 'experimental',
+      ...(resolution === undefined ? {} : { resolution }),
+      chains: strings(isRecord(ids) ? ids.auth_asym_ids : undefined),
     })
-    .sort(
-      (a, b) =>
-        b.coverage - a.coverage ||
-        (a.resolution ?? Infinity) - (b.resolution ?? Infinity),
-    )
+  }
+  return details
+}
+
+/**
+ * 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 function experimentalStructures(
+  alignments: EntityAlignment[],
+  details: ReadonlyMap<string, EntityDetails>,
+): ExperimentalStructure[] {
+  const byEntry = new Map<string, ExperimentalStructure>()
+  for (const a of alignments) {
+    const d = details.get(a.entityId)
+    if (!d) {
+      continue
+    }
+    const existing = byEntry.get(a.pdbId)
+    if (!existing) {
+      byEntry.set(a.pdbId, {
+        pdbId: a.pdbId,
+        method: d.method,
+        ...(d.resolution === undefined ? {} : { resolution: d.resolution }),
+        start: a.start,
+        end: a.end,
+        coverage: a.coverage,
+        chains: [...d.chains],
+      })
+      continue
+    }
+    for (const chain of d.chains) {
+      if (!existing.chains.includes(chain)) {
+        existing.chains.push(chain)
+      }
+    }
+    if (a.coverage > existing.coverage) {
+      Object.assign(existing, {
+        start: a.start,
+        end: a.end,
+        coverage: a.coverage,
+      })
+    }
+  }
+  return [...byEntry.values()].sort(
+    (a, b) =>
+      b.coverage - a.coverage ||
+      (a.resolution ?? Infinity) - (b.resolution ?? Infinity),
+  )
 }
 
-/** Best-effort: an unreachable API reads as "no entries", which costs a
- * structure suggestion and nothing else. */
+const ISOFORM_SUFFIX = /-\d+$/
+
+/**
+ * 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 async function fetchExperimentalStructures(
   uniprotId: string,
   opts?: FetchOptions,
 ): Promise<ExperimentalStructure[]> {
-  for (let attempt = 0; attempt < 2; attempt++) {
-    const res = await rawfetch(beaconsSummaryUrl(uniprotId), opts).catch(
-      () => undefined,
+  const alignments = parseEntityAlignments(
+    await graphqlFetch(
+      RCSB_SEQUENCE_COORDINATES_URL,
+      entityAlignmentsQuery(uniprotId.replace(ISOFORM_SUFFIX, '')),
+      undefined,
+      opts,
+    ),
+  )
+  const ids = alignments.map(a => a.entityId)
+  const details = new Map<string, EntityDetails>()
+  for (let i = 0; i < ids.length; i += DETAILS_BATCH) {
+    const batch = parseEntityDetails(
+      await graphqlFetch(
+        RCSB_DATA_GRAPHQL_URL,
+        ENTITY_DETAILS_QUERY,
+        { ids: ids.slice(i, i + DETAILS_BATCH) },
+        opts,
+      ),
     )
-    if (res?.ok) {
-      return parseExperimentalStructures(await res.json())
-    }
-    if (res?.status === 404 || opts?.signal?.aborted) {
-      return []
+    for (const [id, d] of batch) {
+      details.set(id, d)
     }
   }
-  return []
+  return experimentalStructures(alignments, details)
 }

package/src/structureUrls.ts

@@ -37,6 +37,17 @@ export function getPdbStructureUrl(pdbId: string) {
   return `https://files.rcsb.org/download/${pdbId}.cif`
 }
 
+// A PDB id is four characters, the first a digit: 1TUP, 6ZIO. Checked before
+// a typed one reaches a fetch, so three characters on the way to four are not
+// a 404 apiece.
+export function isPdbId(value: string) {
+  return /^[1-9][A-Za-z0-9]{3}$/.test(value)
+}
+
+export function rcsbEntryUrl(pdbId: string) {
+  return `https://www.rcsb.org/structure/${pdbId.toUpperCase()}`
+}
+
 /**
  * Resolve the `{ pdbId }` shorthand to a concrete structure URL, so a
  * hand-authored snapshot or a short launch URL doesn't have to know RCSB's

package/dist/pdbeBestStructures.d.ts

@@ -1,23 +0,0 @@
-export interface PdbStructureEntry {
-    pdbId: string;
-    experimentalMethod: string;
-    /** Å; absent for NMR and some EM entries */
-    resolution?: number;
-    /** 1-based inclusive UniProt positions the entry covers */
-    unpStart: number;
-    unpEnd: number;
-    /** share of the UniProt sequence the entry covers, 0..1 */
-    coverage: number;
-    chains: string[];
-}
-export declare function pdbeBestStructuresUrl(uniprotId: string): string;
-export declare function isPdbId(value: string): boolean;
-export declare function rcsbEntryUrl(pdbId: string): string;
-/**
- * Groups the per-chain rows into one entry per PDB id, keeping PDBe's order.
- * An entry whose chains cover different UniProt ranges (a construct
- * crystallised beside a different fragment of itself) keeps its first row's
- * range, which is the one PDBe ranked it by. The response is keyed by
- * accession, so the single entry is taken whatever its key.
- */
-export declare function parseBestStructures(json: unknown): PdbStructureEntry[];

package/dist/pdbeBestStructures.js

@@ -1,79 +0,0 @@
-"use strict";
-// PDBe's SIFTS-derived list of the experimental structures covering a UniProt
-// entry, one row per chain, ranked by PDBe (coverage, then resolution). The
-// dialog shows one row per entry, so the chains of a homomer collapse into it.
-Object.defineProperty(exports, "__esModule", { value: true });
-exports.pdbeBestStructuresUrl = pdbeBestStructuresUrl;
-exports.isPdbId = isPdbId;
-exports.rcsbEntryUrl = rcsbEntryUrl;
-exports.parseBestStructures = parseBestStructures;
-function pdbeBestStructuresUrl(uniprotId) {
-    return `https://www.ebi.ac.uk/pdbe/api/mappings/best_structures/${uniprotId}`;
-}
-// A PDB id is four characters, the first a digit: 1TUP, 6ZIO. Checked before
-// a typed one reaches a fetch, so three characters on the way to four are not
-// a 404 apiece.
-function isPdbId(value) {
-    return /^[1-9][A-Za-z0-9]{3}$/.test(value);
-}
-function rcsbEntryUrl(pdbId) {
-    return `https://www.rcsb.org/structure/${pdbId.toUpperCase()}`;
-}
-function isRecord(value) {
-    return typeof value === 'object' && value !== null && !Array.isArray(value);
-}
-function finiteNumber(value) {
-    return typeof value === 'number' && Number.isFinite(value) ? value : undefined;
-}
-/**
- * Groups the per-chain rows into one entry per PDB id, keeping PDBe's order.
- * An entry whose chains cover different UniProt ranges (a construct
- * crystallised beside a different fragment of itself) keeps its first row's
- * range, which is the one PDBe ranked it by. The response is keyed by
- * accession, so the single entry is taken whatever its key.
- */
-function parseBestStructures(json) {
-    const rows = isRecord(json) ? Object.values(json)[0] : undefined;
-    if (!Array.isArray(rows)) {
-        return [];
-    }
-    const byId = new Map();
-    for (const row of rows) {
-        if (!isRecord(row)) {
-            continue;
-        }
-        const pdbId = typeof row.pdb_id === 'string' ? row.pdb_id : undefined;
-        const chain = typeof row.chain_id === 'string' ? row.chain_id : undefined;
-        const unpStart = finiteNumber(row.unp_start);
-        const unpEnd = finiteNumber(row.unp_end);
-        const coverage = finiteNumber(row.coverage);
-        if (pdbId === undefined ||
-            chain === undefined ||
-            unpStart === undefined ||
-            unpEnd === undefined ||
-            coverage === undefined) {
-            continue;
-        }
-        const existing = byId.get(pdbId);
-        if (existing) {
-            if (!existing.chains.includes(chain)) {
-                existing.chains.push(chain);
-            }
-            continue;
-        }
-        const resolution = finiteNumber(row.resolution);
-        byId.set(pdbId, {
-            pdbId,
-            experimentalMethod: typeof row.experimental_method === 'string'
-                ? row.experimental_method
-                : 'unknown',
-            ...(resolution === undefined ? {} : { resolution }),
-            unpStart,
-            unpEnd,
-            coverage,
-            chains: [chain],
-        });
-    }
-    return [...byId.values()];
-}
-//# sourceMappingURL=pdbeBestStructures.js.map

package/dist/pdbeBestStructures.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"pdbeBestStructures.js","sourceRoot":"","sources":["../src/pdbeBestStructures.ts"],"names":[],"mappings":";AAAA,8EAA8E;AAC9E,4EAA4E;AAC5E,+EAA+E;;AAe/E,sDAEC;AAKD,0BAEC;AAED,oCAEC;AAiBD,kDA8CC;AA5ED,SAAgB,qBAAqB,CAAC,SAAiB;IACrD,OAAO,2DAA2D,SAAS,EAAE,CAAA;AAC/E,CAAC;AAED,6EAA6E;AAC7E,8EAA8E;AAC9E,gBAAgB;AAChB,SAAgB,OAAO,CAAC,KAAa;IACnC,OAAO,uBAAuB,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;AAC5C,CAAC;AAED,SAAgB,YAAY,CAAC,KAAa;IACxC,OAAO,kCAAkC,KAAK,CAAC,WAAW,EAAE,EAAE,CAAA;AAChE,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;;;;;;GAMG;AACH,SAAgB,mBAAmB,CAAC,IAAa;IAC/C,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAA;IAChE,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC;QACzB,OAAO,EAAE,CAAA;IACX,CAAC;IACD,MAAM,IAAI,GAAG,IAAI,GAAG,EAA6B,CAAA;IACjD,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACvB,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC;YACnB,SAAQ;QACV,CAAC;QACD,MAAM,KAAK,GAAG,OAAO,GAAG,CAAC,MAAM,KAAK,QAAQ,CAAC,CAAC,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,SAAS,CAAA;QACrE,MAAM,KAAK,GAAG,OAAO,GAAG,CAAC,QAAQ,KAAK,QAAQ,CAAC,CAAC,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,SAAS,CAAA;QACzE,MAAM,QAAQ,GAAG,YAAY,CAAC,GAAG,CAAC,SAAS,CAAC,CAAA;QAC5C,MAAM,MAAM,GAAG,YAAY,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;QACxC,MAAM,QAAQ,GAAG,YAAY,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAA;QAC3C,IACE,KAAK,KAAK,SAAS;YACnB,KAAK,KAAK,SAAS;YACnB,QAAQ,KAAK,SAAS;YACtB,MAAM,KAAK,SAAS;YACpB,QAAQ,KAAK,SAAS,EACtB,CAAC;YACD,SAAQ;QACV,CAAC;QACD,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,CAAA;QAChC,IAAI,QAAQ,EAAE,CAAC;YACb,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;gBACrC,QAAQ,CAAC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;YAC7B,CAAC;YACD,SAAQ;QACV,CAAC;QACD,MAAM,UAAU,GAAG,YAAY,CAAC,GAAG,CAAC,UAAU,CAAC,CAAA;QAC/C,IAAI,CAAC,GAAG,CAAC,KAAK,EAAE;YACd,KAAK;YACL,kBAAkB,EAChB,OAAO,GAAG,CAAC,mBAAmB,KAAK,QAAQ;gBACzC,CAAC,CAAC,GAAG,CAAC,mBAAmB;gBACzB,CAAC,CAAC,SAAS;YACf,GAAG,CAAC,UAAU,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,UAAU,EAAE,CAAC;YACnD,QAAQ;YACR,MAAM;YACN,QAAQ;YACR,MAAM,EAAE,CAAC,KAAK,CAAC;SAChB,CAAC,CAAA;IACJ,CAAC;IACD,OAAO,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC,CAAA;AAC3B,CAAC"}

package/dist/pdbeEntryMolecules.d.ts

@@ -1,7 +0,0 @@
-export declare function pdbeEntryMoleculesUrl(pdbId: string): string;
-/**
- * One sequence per protein entity, in entity order. Nucleic-acid entities are
- * left out: A, C, G, T and U are amino-acid letters too, so a DNA chain scores
- * as a candidate rather than losing honestly.
- */
-export declare function parseEntryMolecules(json: unknown): string[];

package/dist/pdbeEntryMolecules.js

@@ -1,39 +0,0 @@
-"use strict";
-// The polymer entities of a PDB entry, as PDBe already knows them. The dialog
-// wants the entry's residues only to annotate the isoform list, and reading
-// them out of the structure file meant downloading and parsing the whole
-// mmCIF — minutes for a ribosome, for a label. This is a few kB of JSON, and
-// the view downloads the file itself when the user actually launches.
-Object.defineProperty(exports, "__esModule", { value: true });
-exports.pdbeEntryMoleculesUrl = pdbeEntryMoleculesUrl;
-exports.parseEntryMolecules = parseEntryMolecules;
-function pdbeEntryMoleculesUrl(pdbId) {
-    return `https://www.ebi.ac.uk/pdbe/api/pdb/entry/molecules/${pdbId.toLowerCase()}`;
-}
-function isRecord(value) {
-    return typeof value === 'object' && value !== null && !Array.isArray(value);
-}
-/**
- * One sequence per protein entity, in entity order. Nucleic-acid entities are
- * left out: A, C, G, T and U are amino-acid letters too, so a DNA chain scores
- * as a candidate rather than losing honestly.
- */
-function parseEntryMolecules(json) {
-    const entities = isRecord(json) ? Object.values(json)[0] : undefined;
-    if (!Array.isArray(entities)) {
-        return [];
-    }
-    return entities.flatMap(entity => {
-        if (!isRecord(entity)) {
-            return [];
-        }
-        const { molecule_type: type, sequence } = entity;
-        return typeof type === 'string' &&
-            type.startsWith('polypeptide') &&
-            typeof sequence === 'string' &&
-            sequence.length > 0
-            ? [sequence]
-            : [];
-    });
-}
-//# sourceMappingURL=pdbeEntryMolecules.js.map

package/dist/pdbeEntryMolecules.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"pdbeEntryMolecules.js","sourceRoot":"","sources":["../src/pdbeEntryMolecules.ts"],"names":[],"mappings":";AAAA,8EAA8E;AAC9E,4EAA4E;AAC5E,yEAAyE;AACzE,6EAA6E;AAC7E,sEAAsE;;AAEtE,sDAEC;AAWD,kDAiBC;AA9BD,SAAgB,qBAAqB,CAAC,KAAa;IACjD,OAAO,sDAAsD,KAAK,CAAC,WAAW,EAAE,EAAE,CAAA;AACpF,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;;;;GAIG;AACH,SAAgB,mBAAmB,CAAC,IAAa;IAC/C,MAAM,QAAQ,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,SAAS,CAAA;IACpE,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,QAAQ,CAAC,EAAE,CAAC;QAC7B,OAAO,EAAE,CAAA;IACX,CAAC;IACD,OAAO,QAAQ,CAAC,OAAO,CAAC,MAAM,CAAC,EAAE;QAC/B,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;YACtB,OAAO,EAAE,CAAA;QACX,CAAC;QACD,MAAM,EAAE,aAAa,EAAE,IAAI,EAAE,QAAQ,EAAE,GAAG,MAAM,CAAA;QAChD,OAAO,OAAO,IAAI,KAAK,QAAQ;YAC7B,IAAI,CAAC,UAAU,CAAC,aAAa,CAAC;YAC9B,OAAO,QAAQ,KAAK,QAAQ;YAC5B,QAAQ,CAAC,MAAM,GAAG,CAAC;YACnB,CAAC,CAAC,CAAC,QAAQ,CAAC;YACZ,CAAC,CAAC,EAAE,CAAA;IACR,CAAC,CAAC,CAAA;AACJ,CAAC"}

package/dist/siftsNumbering.d.ts

@@ -1,23 +0,0 @@
-import type { UniProtStructureSegment } from './pdbUniProtMapping.ts';
-export interface AuthorRange {
-    start: number;
-    end: number;
-    chain: string;
-    /** author number minus UniProt number in the chosen segment */
-    shift: number;
-}
-/** The segments SIFTS gives one accession, chain ids and author numbers
- * included, out of a whole entry's mappings. */
-export declare function segmentsForAccession(mappings: readonly {
-    accession: string;
-    segments: UniProtStructureSegment[];
-}[], accession: string): UniProtStructureSegment[];
-/**
- * The author-numbered range a UniProt range is cited by in the entry: the
- * segment overlapping most of the range decides the shift. Undefined when no
- * chain of the entry covers any of it, or when SIFTS named no author numbers.
- */
-export declare function toAuthorRange(segments: readonly UniProtStructureSegment[], range: {
-    start: number;
-    end: number;
-}): AuthorRange | undefined;

package/dist/siftsNumbering.js

@@ -1,44 +0,0 @@
-"use strict";
-// How a PDB entry numbers the residues of a UniProt sequence. Author numbering
-// is what a paper cites and what a viewer lights a named residue by, and it is
-// not the UniProt numbering for every entry: haemoglobin's chains number from
-// the mature protein, so author residue 1 is UniProt residue 2 (2HHB, measured
-// 2026-09-12). SIFTS records the correspondence per chain segment, and
-// `fetchUniProtStructureMappings` is what asks PDBe for it.
-Object.defineProperty(exports, "__esModule", { value: true });
-exports.segmentsForAccession = segmentsForAccession;
-exports.toAuthorRange = toAuthorRange;
-/** The segments SIFTS gives one accession, chain ids and author numbers
- * included, out of a whole entry's mappings. */
-function segmentsForAccession(mappings, accession) {
-    return mappings.find(m => m.accession === accession)?.segments ?? [];
-}
-/**
- * The author-numbered range a UniProt range is cited by in the entry: the
- * segment overlapping most of the range decides the shift. Undefined when no
- * chain of the entry covers any of it, or when SIFTS named no author numbers.
- */
-function toAuthorRange(segments, range) {
-    const overlap = (s) => Math.min(s.unpEnd, range.end) - Math.max(s.unpStart, range.start) + 1;
-    const best = segments
-        .flatMap(s => {
-        const { chainId, authorStart, unpStart, unpEnd } = s;
-        return chainId !== undefined &&
-            authorStart !== undefined &&
-            overlap(s) > 0
-            ? [{ chainId, authorStart, unpStart, unpEnd, overlap: overlap(s) }]
-            : [];
-    })
-        .toSorted((a, b) => b.overlap - a.overlap)[0];
-    if (!best) {
-        return undefined;
-    }
-    const shift = best.authorStart - best.unpStart;
-    return {
-        start: Math.max(range.start, best.unpStart) + shift,
-        end: Math.min(range.end, best.unpEnd) + shift,
-        chain: best.chainId,
-        shift,
-    };
-}
-//# sourceMappingURL=siftsNumbering.js.map

package/dist/siftsNumbering.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"siftsNumbering.js","sourceRoot":"","sources":["../src/siftsNumbering.ts"],"names":[],"mappings":";AAAA,+EAA+E;AAC/E,+EAA+E;AAC/E,8EAA8E;AAC9E,+EAA+E;AAC/E,uEAAuE;AACvE,4DAA4D;;AAc5D,oDAQC;AAOD,sCA0BC;AA3CD;gDACgD;AAChD,SAAgB,oBAAoB,CAClC,QAGG,EACH,SAAiB;IAEjB,OAAO,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,SAAS,KAAK,SAAS,CAAC,EAAE,QAAQ,IAAI,EAAE,CAAA;AACtE,CAAC;AAED;;;;GAIG;AACH,SAAgB,aAAa,CAC3B,QAA4C,EAC5C,KAAqC;IAErC,MAAM,OAAO,GAAG,CAAC,CAA0B,EAAE,EAAE,CAC7C,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,MAAM,EAAE,KAAK,CAAC,GAAG,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,QAAQ,EAAE,KAAK,CAAC,KAAK,CAAC,GAAG,CAAC,CAAA;IACvE,MAAM,IAAI,GAAG,QAAQ;SAClB,OAAO,CAAC,CAAC,CAAC,EAAE;QACX,MAAM,EAAE,OAAO,EAAE,WAAW,EAAE,QAAQ,EAAE,MAAM,EAAE,GAAG,CAAC,CAAA;QACpD,OAAO,OAAO,KAAK,SAAS;YAC1B,WAAW,KAAK,SAAS;YACzB,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC;YACd,CAAC,CAAC,CAAC,EAAE,OAAO,EAAE,WAAW,EAAE,QAAQ,EAAE,MAAM,EAAE,OAAO,EAAE,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC;YACnE,CAAC,CAAC,EAAE,CAAA;IACR,CAAC,CAAC;SACD,QAAQ,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,GAAG,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,CAAA;IAC/C,IAAI,CAAC,IAAI,EAAE,CAAC;QACV,OAAO,SAAS,CAAA;IAClB,CAAC;IACD,MAAM,KAAK,GAAG,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,QAAQ,CAAA;IAC9C,OAAO;QACL,KAAK,EAAE,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,KAAK,EAAE,IAAI,CAAC,QAAQ,CAAC,GAAG,KAAK;QACnD,GAAG,EAAE,IAAI,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,EAAE,IAAI,CAAC,MAAM,CAAC,GAAG,KAAK;QAC7C,KAAK,EAAE,IAAI,CAAC,OAAO;QACnB,KAAK;KACN,CAAA;AACH,CAAC"}

package/esm/pdbeBestStructures.d.ts

@@ -1,23 +0,0 @@
-export interface PdbStructureEntry {
-    pdbId: string;
-    experimentalMethod: string;
-    /** Å; absent for NMR and some EM entries */
-    resolution?: number;
-    /** 1-based inclusive UniProt positions the entry covers */
-    unpStart: number;
-    unpEnd: number;
-    /** share of the UniProt sequence the entry covers, 0..1 */
-    coverage: number;
-    chains: string[];
-}
-export declare function pdbeBestStructuresUrl(uniprotId: string): string;
-export declare function isPdbId(value: string): boolean;
-export declare function rcsbEntryUrl(pdbId: string): string;
-/**
- * Groups the per-chain rows into one entry per PDB id, keeping PDBe's order.
- * An entry whose chains cover different UniProt ranges (a construct
- * crystallised beside a different fragment of itself) keeps its first row's
- * range, which is the one PDBe ranked it by. The response is keyed by
- * accession, so the single entry is taken whatever its key.
- */
-export declare function parseBestStructures(json: unknown): PdbStructureEntry[];

package/esm/pdbeBestStructures.js

@@ -1,73 +0,0 @@
-// PDBe's SIFTS-derived list of the experimental structures covering a UniProt
-// entry, one row per chain, ranked by PDBe (coverage, then resolution). The
-// dialog shows one row per entry, so the chains of a homomer collapse into it.
-export function pdbeBestStructuresUrl(uniprotId) {
-    return `https://www.ebi.ac.uk/pdbe/api/mappings/best_structures/${uniprotId}`;
-}
-// A PDB id is four characters, the first a digit: 1TUP, 6ZIO. Checked before
-// a typed one reaches a fetch, so three characters on the way to four are not
-// a 404 apiece.
-export function isPdbId(value) {
-    return /^[1-9][A-Za-z0-9]{3}$/.test(value);
-}
-export function rcsbEntryUrl(pdbId) {
-    return `https://www.rcsb.org/structure/${pdbId.toUpperCase()}`;
-}
-function isRecord(value) {
-    return typeof value === 'object' && value !== null && !Array.isArray(value);
-}
-function finiteNumber(value) {
-    return typeof value === 'number' && Number.isFinite(value) ? value : undefined;
-}
-/**
- * Groups the per-chain rows into one entry per PDB id, keeping PDBe's order.
- * An entry whose chains cover different UniProt ranges (a construct
- * crystallised beside a different fragment of itself) keeps its first row's
- * range, which is the one PDBe ranked it by. The response is keyed by
- * accession, so the single entry is taken whatever its key.
- */
-export function parseBestStructures(json) {
-    const rows = isRecord(json) ? Object.values(json)[0] : undefined;
-    if (!Array.isArray(rows)) {
-        return [];
-    }
-    const byId = new Map();
-    for (const row of rows) {
-        if (!isRecord(row)) {
-            continue;
-        }
-        const pdbId = typeof row.pdb_id === 'string' ? row.pdb_id : undefined;
-        const chain = typeof row.chain_id === 'string' ? row.chain_id : undefined;
-        const unpStart = finiteNumber(row.unp_start);
-        const unpEnd = finiteNumber(row.unp_end);
-        const coverage = finiteNumber(row.coverage);
-        if (pdbId === undefined ||
-            chain === undefined ||
-            unpStart === undefined ||
-            unpEnd === undefined ||
-            coverage === undefined) {
-            continue;
-        }
-        const existing = byId.get(pdbId);
-        if (existing) {
-            if (!existing.chains.includes(chain)) {
-                existing.chains.push(chain);
-            }
-            continue;
-        }
-        const resolution = finiteNumber(row.resolution);
-        byId.set(pdbId, {
-            pdbId,
-            experimentalMethod: typeof row.experimental_method === 'string'
-                ? row.experimental_method
-                : 'unknown',
-            ...(resolution === undefined ? {} : { resolution }),
-            unpStart,
-            unpEnd,
-            coverage,
-            chains: [chain],
-        });
-    }
-    return [...byId.values()];
-}
-//# sourceMappingURL=pdbeBestStructures.js.map