model.ts 17 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430
  1. /**
  2. * Copyright (c) 2017-2023 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 { UUID } from '../../../mol-util/uuid';
  8. import { StructureSequence } from './properties/sequence';
  9. import { AtomicHierarchy, AtomicConformation, AtomicRanges } from './properties/atomic';
  10. import { CoarseHierarchy, CoarseConformation } from './properties/coarse';
  11. import { Entities, ChemicalComponentMap, MissingResidues, StructAsymMap } from './properties/common';
  12. import { CustomProperties } from '../../custom-property';
  13. import { SaccharideComponentMap } from '../structure/carbohydrates/constants';
  14. import { ModelFormat } from '../../../mol-model-formats/format';
  15. import { calcModelCenter, getAsymIdCount } from './util';
  16. import { Vec3 } from '../../../mol-math/linear-algebra';
  17. import { Coordinates } from '../coordinates';
  18. import { Topology } from '../topology';
  19. import { Task } from '../../../mol-task';
  20. import { IndexPairBonds } from '../../../mol-model-formats/structure/property/bonds/index-pair';
  21. import { createModels } from '../../../mol-model-formats/structure/basic/parser';
  22. import { MmcifFormat } from '../../../mol-model-formats/structure/mmcif';
  23. import { ChainIndex, ElementIndex } from './indexing';
  24. import { SymmetryOperator } from '../../../mol-math/geometry';
  25. import { ModelSymmetry } from '../../../mol-model-formats/structure/property/symmetry';
  26. import { Column } from '../../../mol-data/db';
  27. import { CustomModelProperty } from '../../../mol-model-props/common/custom-model-property';
  28. import { Trajectory, ArrayTrajectory } from '../trajectory';
  29. import { Unit } from '../structure';
  30. import { SortedArray } from '../../../mol-data/int/sorted-array';
  31. import { PolymerType } from './types';
  32. import { ModelSecondaryStructure } from '../../../mol-model-formats/structure/property/secondary-structure';
  33. /**
  34. * Interface to the "source data" of the molecule.
  35. *
  36. * "Atoms" are integers in the range [0, atomCount).
  37. */
  38. export interface Model extends Readonly<{
  39. id: UUID,
  40. entryId: string,
  41. label: string,
  42. /** the name of the entry/file/collection the model is part of */
  43. entry: string,
  44. /**
  45. * corresponds to
  46. * - for IHM: `ihm_model_list.model_id`
  47. * - for standard mmCIF: `atom_site.pdbx_PDB_model_num`
  48. * - for models from coordinates: frame index
  49. */
  50. modelNum: number,
  51. sourceData: ModelFormat,
  52. parent: Model | undefined,
  53. entities: Entities,
  54. sequence: StructureSequence,
  55. atomicHierarchy: AtomicHierarchy,
  56. atomicConformation: AtomicConformation,
  57. atomicRanges: AtomicRanges,
  58. atomicChainOperatorMappinng: Map<ChainIndex, SymmetryOperator>,
  59. properties: {
  60. /** map that holds details about unobserved or zero occurrence residues */
  61. readonly missingResidues: MissingResidues,
  62. /** maps residue name to `ChemicalComponent` data */
  63. readonly chemicalComponentMap: ChemicalComponentMap
  64. /** maps residue name to `SaccharideComponent` data */
  65. readonly saccharideComponentMap: SaccharideComponentMap
  66. /** maps label_asym_id name to `StructAsym` data */
  67. readonly structAsymMap: StructAsymMap
  68. },
  69. customProperties: CustomProperties,
  70. /**
  71. * Not to be accessed directly, each custom property descriptor
  72. * defines property accessors that use this field to store the data.
  73. */
  74. _staticPropertyData: { [name: string]: any },
  75. _dynamicPropertyData: { [name: string]: any },
  76. coarseHierarchy: CoarseHierarchy,
  77. coarseConformation: CoarseConformation
  78. }> {
  79. } { }
  80. export namespace Model {
  81. function _trajectoryFromModelAndCoordinates(model: Model, coordinates: Coordinates) {
  82. const trajectory: Model[] = [];
  83. const { frames } = coordinates;
  84. const srcIndex = model.atomicHierarchy.atomSourceIndex;
  85. const isIdentity = Column.isIdentity(srcIndex);
  86. const srcIndexArray = isIdentity ? void 0 : srcIndex.toArray({ array: Int32Array });
  87. const coarseGrained = isCoarseGrained(model);
  88. const elementCount = model.atomicHierarchy.atoms._rowCount;
  89. for (let i = 0, il = frames.length; i < il; ++i) {
  90. const f = frames[i];
  91. if (f.elementCount !== elementCount) {
  92. throw new Error(`Frame element count mismatch, got ${f.elementCount} but expected ${elementCount}.`);
  93. }
  94. const m = {
  95. ...model,
  96. id: UUID.create22(),
  97. modelNum: i,
  98. atomicConformation: Coordinates.getAtomicConformation(f, {
  99. atomId: model.atomicConformation.atomId,
  100. occupancy: model.atomicConformation.occupancy,
  101. B_iso_or_equiv: model.atomicConformation.B_iso_or_equiv
  102. }, srcIndexArray),
  103. // TODO: add support for supplying sphere and gaussian coordinates in addition to atomic coordinates?
  104. // coarseConformation: coarse.conformation,
  105. customProperties: new CustomProperties(),
  106. _staticPropertyData: Object.create(null),
  107. _dynamicPropertyData: Object.create(null)
  108. };
  109. if (f.cell) {
  110. const symmetry = ModelSymmetry.fromCell(f.cell.size, f.cell.anglesInRadians);
  111. ModelSymmetry.Provider.set(m, symmetry);
  112. }
  113. TrajectoryInfo.set(m, { index: i, size: frames.length });
  114. CoarseGrained.set(m, coarseGrained);
  115. trajectory.push(m);
  116. }
  117. return { trajectory, srcIndexArray };
  118. }
  119. export function trajectoryFromModelAndCoordinates(model: Model, coordinates: Coordinates): Trajectory {
  120. return new ArrayTrajectory(_trajectoryFromModelAndCoordinates(model, coordinates).trajectory);
  121. }
  122. export function trajectoryFromTopologyAndCoordinates(topology: Topology, coordinates: Coordinates): Task<Trajectory> {
  123. return Task.create('Create Trajectory', async ctx => {
  124. const models = await createModels(topology.basic, topology.sourceData, ctx);
  125. if (models.frameCount === 0) throw new Error('found no model');
  126. const model = models.representative;
  127. const { trajectory } = _trajectoryFromModelAndCoordinates(model, coordinates);
  128. const bondData = { pairs: topology.bonds, count: model.atomicHierarchy.atoms._rowCount };
  129. const indexPairBonds = IndexPairBonds.fromData(bondData);
  130. const coarseGrained = isCoarseGrained(model);
  131. let index = 0;
  132. for (const m of trajectory) {
  133. IndexPairBonds.Provider.set(m, indexPairBonds);
  134. TrajectoryInfo.set(m, { index: index++, size: trajectory.length });
  135. CoarseGrained.set(m, coarseGrained);
  136. }
  137. return new ArrayTrajectory(trajectory);
  138. });
  139. }
  140. const CenterProp = '__Center__';
  141. export function getCenter(model: Model): Vec3 {
  142. if (model._dynamicPropertyData[CenterProp]) return model._dynamicPropertyData[CenterProp];
  143. const center = calcModelCenter(model.atomicConformation, model.coarseConformation);
  144. model._dynamicPropertyData[CenterProp] = center;
  145. return center;
  146. }
  147. function invertIndex(xs: Column<number>) {
  148. const invertedIndex = new Int32Array(xs.rowCount);
  149. let isIdentity = false;
  150. for (let i = 0, _i = xs.rowCount; i < _i; i++) {
  151. const x = xs.value(i);
  152. if (x !== i) isIdentity = false;
  153. invertedIndex[x] = i;
  154. }
  155. return { isIdentity, invertedIndex: invertedIndex as unknown as ArrayLike<ElementIndex> };
  156. }
  157. const InvertedAtomSrcIndexProp = '__InvertedAtomSrcIndex__';
  158. export function getInvertedAtomSourceIndex(model: Model): { isIdentity: boolean, invertedIndex: ArrayLike<ElementIndex> } {
  159. if (model._staticPropertyData[InvertedAtomSrcIndexProp]) return model._staticPropertyData[InvertedAtomSrcIndexProp];
  160. const index = invertIndex(model.atomicHierarchy.atomSourceIndex);
  161. model._staticPropertyData[InvertedAtomSrcIndexProp] = index;
  162. return index;
  163. }
  164. const TrajectoryInfoProp = '__TrajectoryInfo__';
  165. export type TrajectoryInfo = { readonly index: number, readonly size: number }
  166. export const TrajectoryInfo = {
  167. get(model: Model): TrajectoryInfo {
  168. return model._dynamicPropertyData[TrajectoryInfoProp] || { index: 0, size: 1 };
  169. },
  170. set(model: Model, trajectoryInfo: TrajectoryInfo) {
  171. return model._dynamicPropertyData[TrajectoryInfoProp] = trajectoryInfo;
  172. }
  173. };
  174. const AsymIdCountProp = '__AsymIdCount__';
  175. export type AsymIdCount = { readonly auth: number, readonly label: number }
  176. export const AsymIdCount = {
  177. get(model: Model): AsymIdCount {
  178. if (model._dynamicPropertyData[AsymIdCountProp]) return model._dynamicPropertyData[AsymIdCountProp];
  179. const asymIdCount = getAsymIdCount(model);
  180. model._dynamicPropertyData[AsymIdCountProp] = asymIdCount;
  181. return asymIdCount;
  182. },
  183. };
  184. export type AsymIdOffset = { auth: number, label: number };
  185. export const AsymIdOffset = CustomModelProperty.createSimple<AsymIdOffset>('asym_id_offset', 'static');
  186. export type Index = number;
  187. export const Index = CustomModelProperty.createSimple<Index>('index', 'static');
  188. export type MaxIndex = number;
  189. export const MaxIndex = CustomModelProperty.createSimple<MaxIndex>('max_index', 'static');
  190. export function getRoot(model: Model) {
  191. return model.parent || model;
  192. }
  193. export function areHierarchiesEqual(a: Model, b: Model) {
  194. return a.atomicHierarchy === b.atomicHierarchy && a.coarseHierarchy === b.coarseHierarchy;
  195. }
  196. const CoordinatesHistoryProp = '__CoordinatesHistory__';
  197. export type CoordinatesHistory = {
  198. areEqual(elements: SortedArray<ElementIndex>, kind: Unit.Kind, model: Model): boolean
  199. }
  200. export const CoordinatesHistory = {
  201. get(model: Model): CoordinatesHistory | undefined {
  202. return model._staticPropertyData[CoordinatesHistoryProp];
  203. },
  204. set(model: Model, coordinatesHistory: CoordinatesHistory) {
  205. return model._staticPropertyData[CoordinatesHistoryProp] = coordinatesHistory;
  206. }
  207. };
  208. const CoarseGrainedProp = '__CoarseGrained__';
  209. export const CoarseGrained = {
  210. get(model: Model): boolean | undefined {
  211. return model._staticPropertyData[CoarseGrainedProp];
  212. },
  213. set(model: Model, coarseGrained: boolean) {
  214. return model._staticPropertyData[CoarseGrainedProp] = coarseGrained;
  215. }
  216. };
  217. /**
  218. * Has typical coarse grained atom names (BB, SC1) or less than three times as many
  219. * atoms as polymer residues (C-alpha only models).
  220. */
  221. export function isCoarseGrained(model: Model): boolean {
  222. let coarseGrained = CoarseGrained.get(model);
  223. if (coarseGrained === undefined) {
  224. let polymerResidueCount = 0;
  225. const { polymerType } = model.atomicHierarchy.derived.residue;
  226. for (let i = 0; i < polymerType.length; ++i) {
  227. if (polymerType[i] !== PolymerType.NA) polymerResidueCount += 1;
  228. }
  229. // check for coarse grained atom names
  230. let hasBB = false, hasSC1 = false;
  231. const { label_atom_id, _rowCount: atomCount } = model.atomicHierarchy.atoms;
  232. for (let i = 0; i < atomCount; ++i) {
  233. const atomName = label_atom_id.value(i);
  234. if (!hasBB && atomName === 'BB') hasBB = true;
  235. if (!hasSC1 && atomName === 'SC1') hasSC1 = true;
  236. if (hasBB && hasSC1) break;
  237. }
  238. coarseGrained = (hasBB && hasSC1) || (
  239. polymerResidueCount && atomCount
  240. ? atomCount / polymerResidueCount < 3
  241. : false
  242. );
  243. CoarseGrained.set(model, coarseGrained);
  244. }
  245. return coarseGrained;
  246. }
  247. //
  248. export function hasCarbohydrate(model: Model): boolean {
  249. return model.properties.saccharideComponentMap.size > 0;
  250. }
  251. export function hasProtein(model: Model): boolean {
  252. const { subtype } = model.entities;
  253. for (let i = 0, il = subtype.rowCount; i < il; ++i) {
  254. if (subtype.value(i).startsWith('polypeptide')) return true;
  255. }
  256. return false;
  257. }
  258. export function hasNucleic(model: Model): boolean {
  259. const { subtype } = model.entities;
  260. for (let i = 0, il = subtype.rowCount; i < il; ++i) {
  261. const s = subtype.value(i);
  262. if (s.endsWith('ribonucleotide hybrid') || s.endsWith('ribonucleotide')) return true;
  263. }
  264. return false;
  265. }
  266. export function isFromPdbArchive(model: Model): boolean {
  267. if (!MmcifFormat.is(model.sourceData)) return false;
  268. const { db } = model.sourceData.data;
  269. for (let i = 0, il = db.database_2.database_id.rowCount; i < il; ++i) {
  270. if (db.database_2.database_id.value(i) === 'pdb') return true;
  271. }
  272. return false;
  273. }
  274. export function hasPdbId(model: Model): boolean {
  275. if (!MmcifFormat.is(model.sourceData)) return false;
  276. return (
  277. // 4 character PDB id
  278. model.entryId.match(/^[1-9][a-z0-9]{3,3}$/i) !== null ||
  279. // long PDB id
  280. model.entryId.match(/^pdb_[0-9]{4,4}[1-9][a-z0-9]{3,3}$/i) !== null
  281. );
  282. }
  283. export function hasSecondaryStructure(model: Model): boolean {
  284. if (MmcifFormat.is(model.sourceData)) {
  285. const { db } = model.sourceData.data;
  286. return (
  287. db.struct_conf.id.isDefined ||
  288. db.struct_sheet_range.id.isDefined
  289. );
  290. } else {
  291. return ModelSecondaryStructure.Provider.isApplicable(model);
  292. }
  293. }
  294. const tmpAngles90 = Vec3.create(1.5707963, 1.5707963, 1.5707963); // in radians
  295. const tmpLengths1 = Vec3.create(1, 1, 1);
  296. export function hasCrystalSymmetry(model: Model): boolean {
  297. const spacegroup = ModelSymmetry.Provider.get(model)?.spacegroup;
  298. return !!spacegroup && !(
  299. spacegroup.num === 1 &&
  300. Vec3.equals(spacegroup.cell.anglesInRadians, tmpAngles90) &&
  301. Vec3.equals(spacegroup.cell.size, tmpLengths1)
  302. );
  303. }
  304. export function isFromXray(model: Model): boolean {
  305. if (!MmcifFormat.is(model.sourceData)) return false;
  306. const { db } = model.sourceData.data;
  307. for (let i = 0; i < db.exptl.method.rowCount; i++) {
  308. const v = db.exptl.method.value(i).toUpperCase();
  309. if (v.indexOf('DIFFRACTION') >= 0) return true;
  310. }
  311. return false;
  312. }
  313. export function isFromEm(model: Model): boolean {
  314. if (!MmcifFormat.is(model.sourceData)) return false;
  315. const { db } = model.sourceData.data;
  316. for (let i = 0; i < db.exptl.method.rowCount; i++) {
  317. const v = db.exptl.method.value(i).toUpperCase();
  318. if (v.indexOf('MICROSCOPY') >= 0) return true;
  319. }
  320. return false;
  321. }
  322. export function isFromNmr(model: Model): boolean {
  323. if (!MmcifFormat.is(model.sourceData)) return false;
  324. const { db } = model.sourceData.data;
  325. for (let i = 0; i < db.exptl.method.rowCount; i++) {
  326. const v = db.exptl.method.value(i).toUpperCase();
  327. if (v.indexOf('NMR') >= 0) return true;
  328. }
  329. return false;
  330. }
  331. export function hasXrayMap(model: Model): boolean {
  332. if (!MmcifFormat.is(model.sourceData)) return false;
  333. // Check exprimental method to exclude models solved with
  334. // 'ELECTRON CRYSTALLOGRAPHY' which also have structure factors
  335. if (!isFromXray(model)) return false;
  336. const { db } = model.sourceData.data;
  337. const { status_code_sf } = db.pdbx_database_status;
  338. return status_code_sf.isDefined && status_code_sf.value(0) === 'REL';
  339. }
  340. /**
  341. * Also checks for `content_type` of 'associated EM volume' to exclude cases
  342. * like 6TEK which are solved with 'X-RAY DIFFRACTION' but have an related
  343. * EMDB entry of type 'other EM volume'.
  344. */
  345. export function hasEmMap(model: Model): boolean {
  346. if (!MmcifFormat.is(model.sourceData)) return false;
  347. const { db } = model.sourceData.data;
  348. const { db_name, content_type } = db.pdbx_database_related;
  349. for (let i = 0, il = db.pdbx_database_related._rowCount; i < il; ++i) {
  350. if (db_name.value(i).toUpperCase() === 'EMDB' && content_type.value(i) === 'associated EM volume') {
  351. return true;
  352. }
  353. }
  354. return false;
  355. }
  356. export function hasDensityMap(model: Model): boolean {
  357. if (!MmcifFormat.is(model.sourceData)) return false;
  358. return hasXrayMap(model) || hasEmMap(model);
  359. }
  360. export function probablyHasDensityMap(model: Model): boolean {
  361. if (!MmcifFormat.is(model.sourceData)) return false;
  362. const { db } = model.sourceData.data;
  363. return hasDensityMap(model) || (
  364. // check if from pdb archive but missing relevant meta data
  365. hasPdbId(model) && (
  366. !db.exptl.method.isDefined ||
  367. (isFromXray(model) && (
  368. !db.pdbx_database_status.status_code_sf.isDefined ||
  369. db.pdbx_database_status.status_code_sf.valueKind(0) === Column.ValueKinds.Unknown
  370. )) ||
  371. (isFromEm(model) && (
  372. !db.pdbx_database_related.db_name.isDefined
  373. ))
  374. )
  375. );
  376. }
  377. }