atomic-index.ts 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249
  1. /**
  2. * Copyright (c) 2017-2020 mol* contributors, licensed under MIT, See LICENSE file for more info.
  3. *
  4. * @author David Sehnal <david.sehnal@gmail.com>
  5. * @author Alexander Rose <alexander.rose@weirdbyte.de>
  6. */
  7. import { AtomicData, AtomicSegments } from '../atomic'
  8. import { Interval, Segmentation, SortedArray } from '../../../../../mol-data/int'
  9. import { Entities } from '../common'
  10. import { ChainIndex, ResidueIndex, EntityIndex, ElementIndex } from '../../indexing';
  11. import { AtomicIndex, AtomicHierarchy } from '../atomic/hierarchy';
  12. import { cantorPairing } from '../../../../../mol-data/util';
  13. import { Column } from '../../../../../mol-data/db';
  14. function getResidueId(seq_id: number, ins_code: string) {
  15. if (!ins_code) return seq_id;
  16. if (ins_code.length === 1) return cantorPairing(ins_code.charCodeAt(0), seq_id);
  17. if (ins_code.length === 2) return cantorPairing(ins_code.charCodeAt(0), cantorPairing(ins_code.charCodeAt(1), seq_id));
  18. return `${seq_id} ${ins_code}`;
  19. }
  20. function updateMapMapIndex<K, I extends number>(map: Map<K, Map<string, I>>, key0: K, key1: string, index: I) {
  21. if (map.has(key0)) {
  22. const submap = map.get(key0)!;
  23. if (!submap.has(key1)) {
  24. submap.set(key1, index);
  25. }
  26. } else {
  27. const submap = new Map<string, I>();
  28. map.set(key0, submap);
  29. submap.set(key1, index);
  30. }
  31. }
  32. function missingEntity(k: string) {
  33. throw new Error(`Missing entity entry for entity id '${k}'.`);
  34. }
  35. interface Mapping {
  36. entities: Entities,
  37. label_seq_id: SortedArray,
  38. label_atom_id: Column<string>,
  39. auth_atom_id: Column<string>,
  40. label_alt_id: Column<string>,
  41. segments: AtomicSegments,
  42. chain_index_entity_index: EntityIndex[],
  43. entity_index_label_asym_id: Map<EntityIndex, Map<string, ChainIndex>>,
  44. chain_index_label_seq_id: Map<ChainIndex, Map<string | number, ResidueIndex>>,
  45. auth_asym_id_auth_seq_id: Map<string, Map<number, ChainIndex>>,
  46. chain_index_auth_seq_id: Map<ChainIndex, Map<string | number, ResidueIndex>>,
  47. label_asym_id: Map<string, EntityIndex>,
  48. }
  49. function createMapping(entities: Entities, data: AtomicData, segments: AtomicSegments): Mapping {
  50. return {
  51. entities,
  52. segments,
  53. label_seq_id: SortedArray.ofSortedArray(data.residues.label_seq_id.toArray({ array: Int32Array })),
  54. label_atom_id: data.atoms.label_atom_id,
  55. auth_atom_id: data.atoms.auth_atom_id,
  56. label_alt_id: data.atoms.label_alt_id,
  57. chain_index_entity_index: new Int32Array(data.chains._rowCount) as any,
  58. entity_index_label_asym_id: new Map(),
  59. chain_index_label_seq_id: new Map(),
  60. auth_asym_id_auth_seq_id: new Map(),
  61. chain_index_auth_seq_id: new Map(),
  62. label_asym_id: new Map(),
  63. };
  64. }
  65. const _tempResidueKey = AtomicIndex.EmptyResidueKey();
  66. class Index implements AtomicIndex {
  67. private entityIndex: Entities['getEntityIndex'];
  68. private residueOffsets: ArrayLike<ElementIndex>;
  69. getEntityFromChain(cI: ChainIndex): EntityIndex {
  70. return this.map.chain_index_entity_index[cI];
  71. }
  72. findEntity(label_asym_id: string): EntityIndex {
  73. const entityIndex = this.map.label_asym_id.get(label_asym_id)
  74. return entityIndex !== undefined ? entityIndex : -1 as EntityIndex
  75. }
  76. findChainLabel(key: AtomicIndex.ChainLabelKey): ChainIndex {
  77. const eI = this.entityIndex(key.label_entity_id);
  78. if (eI < 0 || !this.map.entity_index_label_asym_id.has(eI)) return -1 as ChainIndex;
  79. const cm = this.map.entity_index_label_asym_id.get(eI);
  80. if (!cm) return -1 as ChainIndex;
  81. return cm.has(key.label_asym_id) ? cm.get(key.label_asym_id)! : -1 as ChainIndex;
  82. }
  83. findChainAuth(key: AtomicIndex.ChainAuthKey): ChainIndex {
  84. if (!this.map.auth_asym_id_auth_seq_id.has(key.auth_asym_id)) return -1 as ChainIndex
  85. const rm = this.map.auth_asym_id_auth_seq_id.get(key.auth_asym_id)!
  86. return rm.has(key.auth_seq_id) ? rm.get(key.auth_seq_id)! : -1 as ChainIndex
  87. }
  88. findResidue(label_entity_id: string, label_asym_id: string, auth_seq_id: number, pdbx_PDB_ins_code?: string): ResidueIndex
  89. findResidue(key: AtomicIndex.ResidueKey): ResidueIndex
  90. findResidue(label_entity_id_or_key: string | AtomicIndex.ResidueKey, label_asym_id?: string, auth_seq_id?: number, pdbx_PDB_ins_code?: string): ResidueIndex {
  91. let key: AtomicIndex.ResidueKey;
  92. if (arguments.length === 1) {
  93. key = label_entity_id_or_key as AtomicIndex.ResidueKey
  94. } else {
  95. _tempResidueKey.label_entity_id = label_entity_id_or_key as string;
  96. _tempResidueKey.label_asym_id = label_asym_id!;
  97. _tempResidueKey.auth_seq_id = auth_seq_id!;
  98. _tempResidueKey.pdbx_PDB_ins_code = pdbx_PDB_ins_code;
  99. key = _tempResidueKey;
  100. }
  101. const cI = this.findChainLabel(key);
  102. if (cI < 0) return -1 as ResidueIndex;
  103. const rm = this.map.chain_index_auth_seq_id.get(cI)!;
  104. const id = getResidueId(key.auth_seq_id, key.pdbx_PDB_ins_code || '');
  105. return rm.has(id) ? rm.get(id)! : -1 as ResidueIndex;
  106. }
  107. findResidueAuth(key: AtomicIndex.ResidueAuthKey): ResidueIndex {
  108. const cI = this.findChainAuth(key);
  109. if (cI < 0) return -1 as ResidueIndex;
  110. const rm = this.map.chain_index_auth_seq_id.get(cI)!;
  111. const id = getResidueId(key.auth_seq_id, key.pdbx_PDB_ins_code || '');
  112. return rm.has(id) ? rm.get(id)! : -1 as ResidueIndex;
  113. }
  114. findResidueInsertion(key: AtomicIndex.ResidueLabelKey): ResidueIndex {
  115. const cI = this.findChainLabel(key);
  116. if (cI < 0) return -1 as ResidueIndex;
  117. const rm = this.map.chain_index_label_seq_id.get(cI)!;
  118. const id = getResidueId(key.label_seq_id, key.pdbx_PDB_ins_code || '');
  119. if (rm.has(id)) return rm.get(id)!;
  120. const idx = SortedArray.findPredecessorIndex(this.map.label_seq_id, key.label_seq_id) as ResidueIndex;
  121. const start = AtomicHierarchy.chainStartResidueIndex(this.map.segments, cI);
  122. if (idx < start) return start;
  123. const end = AtomicHierarchy.chainEndResidueIndexExcl(this.map.segments, cI) - 1 as ResidueIndex;
  124. if (idx >= end) return end;
  125. return idx;
  126. }
  127. findAtom(key: AtomicIndex.AtomKey): ElementIndex {
  128. const rI = this.findResidue(key);
  129. if (rI < 0) return -1 as ElementIndex;
  130. if (typeof key.label_alt_id === 'undefined') {
  131. return findAtomByName(this.residueOffsets[rI], this.residueOffsets[rI + 1], this.map.label_atom_id, key.label_atom_id);
  132. }
  133. return findAtomByNameAndAltLoc(this.residueOffsets[rI], this.residueOffsets[rI + 1], this.map.label_atom_id, this.map.label_alt_id, key.label_atom_id, key.label_alt_id);
  134. }
  135. findAtomAuth(key: AtomicIndex.AtomAuthKey): ElementIndex {
  136. const rI = this.findResidueAuth(key);
  137. if (rI < 0) return -1 as ElementIndex;
  138. if (typeof key.label_alt_id === 'undefined') {
  139. return findAtomByName(this.residueOffsets[rI], this.residueOffsets[rI + 1], this.map.auth_atom_id, key.auth_atom_id);
  140. }
  141. return findAtomByNameAndAltLoc(this.residueOffsets[rI], this.residueOffsets[rI + 1], this.map.auth_atom_id, this.map.label_alt_id, key.auth_atom_id, key.label_alt_id);
  142. }
  143. findAtomOnResidue(rI: ResidueIndex, label_atom_id: string, label_alt_id?: string) {
  144. if (typeof label_alt_id === 'undefined') {
  145. return findAtomByName(this.residueOffsets[rI], this.residueOffsets[rI + 1], this.map.label_atom_id, label_atom_id);
  146. }
  147. return findAtomByNameAndAltLoc(this.residueOffsets[rI], this.residueOffsets[rI + 1], this.map.label_atom_id, this.map.label_alt_id, label_atom_id, label_alt_id);
  148. }
  149. findAtomsOnResidue(rI: ResidueIndex, label_atom_ids: Set<string>) {
  150. return findAtomByNames(this.residueOffsets[rI], this.residueOffsets[rI + 1], this.map.label_atom_id, label_atom_ids)
  151. }
  152. constructor(private map: Mapping) {
  153. this.entityIndex = map.entities.getEntityIndex;
  154. this.residueOffsets = this.map.segments.residueAtomSegments.offsets;
  155. }
  156. }
  157. function findAtomByName(start: ElementIndex, end: ElementIndex, data: Column<string>, atomName: string): ElementIndex {
  158. for (let i = start; i < end; i++) {
  159. if (data.value(i) === atomName) return i;
  160. }
  161. return -1 as ElementIndex;
  162. }
  163. function findAtomByNames(start: ElementIndex, end: ElementIndex, data: Column<string>, atomNames: Set<string>): ElementIndex {
  164. for (let i = start; i < end; i++) {
  165. if (atomNames.has(data.value(i))) return i;
  166. }
  167. return -1 as ElementIndex;
  168. }
  169. function findAtomByNameAndAltLoc(start: ElementIndex, end: ElementIndex, nameData: Column<string>, altLocData: Column<string>,
  170. atomName: string, altLoc: string): ElementIndex {
  171. for (let i = start; i < end; i++) {
  172. if (nameData.value(i) === atomName && altLocData.value(i) === altLoc) return i;
  173. }
  174. return -1 as ElementIndex;
  175. }
  176. export function getAtomicIndex(data: AtomicData, entities: Entities, segments: AtomicSegments): AtomicIndex {
  177. const map = createMapping(entities, data, segments);
  178. const { label_seq_id, auth_seq_id, pdbx_PDB_ins_code } = data.residues;
  179. const { label_entity_id, label_asym_id, auth_asym_id } = data.chains;
  180. const atomSet = Interval.ofBounds(0, data.atoms._rowCount);
  181. const chainsIt = Segmentation.transientSegments(segments.chainAtomSegments, atomSet);
  182. while (chainsIt.hasNext) {
  183. const chainSegment = chainsIt.move();
  184. const chainIndex = chainSegment.index;
  185. const entityIndex = entities.getEntityIndex(label_entity_id.value(chainIndex));
  186. if (entityIndex < 0) missingEntity(label_entity_id.value(chainIndex));
  187. map.chain_index_entity_index[chainIndex] = entityIndex;
  188. const authAsymId = auth_asym_id.value(chainIndex);
  189. let auth_asym_id_auth_seq_id = map.auth_asym_id_auth_seq_id.get(authAsymId)
  190. if (!auth_asym_id_auth_seq_id) {
  191. auth_asym_id_auth_seq_id = new Map<number, ChainIndex>();
  192. map.auth_asym_id_auth_seq_id.set(authAsymId, auth_asym_id_auth_seq_id)
  193. }
  194. const labelAsymId = label_asym_id.value(chainIndex)
  195. if (!map.label_asym_id.has(labelAsymId)) map.label_asym_id.set(labelAsymId, entityIndex);
  196. updateMapMapIndex(map.entity_index_label_asym_id, entityIndex, labelAsymId, chainIndex);
  197. const chain_index_label_seq_id = new Map<string | number, ResidueIndex>();
  198. const chain_index_auth_seq_id = new Map<string | number, ResidueIndex>();
  199. map.chain_index_label_seq_id.set(chainIndex, chain_index_label_seq_id);
  200. map.chain_index_auth_seq_id.set(chainIndex, chain_index_auth_seq_id);
  201. const residuesIt = Segmentation.transientSegments(segments.residueAtomSegments, atomSet, chainSegment);
  202. while (residuesIt.hasNext) {
  203. const residueSegment = residuesIt.move();
  204. const rI = residueSegment.index;
  205. const authSeqId = auth_seq_id.value(rI)
  206. const insCode = pdbx_PDB_ins_code.value(rI);
  207. chain_index_label_seq_id.set(getResidueId(label_seq_id.value(rI), insCode), rI);
  208. chain_index_auth_seq_id.set(getResidueId(authSeqId, insCode), rI);
  209. auth_asym_id_auth_seq_id.set(authSeqId, chainIndex)
  210. }
  211. }
  212. return new Index(map);
  213. }