From 0b9f31b45e79fd5d3931894cf8e36ce1382cc7f1 Mon Sep 17 00:00:00 2001
From: David Sehnal <david.sehnal@gmail.com>
Date: Fri, 8 Jun 2018 12:58:51 +0200
Subject: [PATCH] Mod res support
---
src/apps/structure-info/model.ts | 12 ++++
.../structure/model/formats/mmcif.ts | 17 ++++-
.../model/formats/mmcif/modified-residues.ts | 65 +++++++++++++++++++
src/mol-model/structure/model/model.ts | 7 +-
.../model/properties/atomic/hierarchy.ts | 4 +-
5 files changed, 101 insertions(+), 4 deletions(-)
create mode 100644 src/mol-model/structure/model/formats/mmcif/modified-residues.ts
diff --git a/src/apps/structure-info/model.ts b/src/apps/structure-info/model.ts
index 85f118dea..384758e57 100644
--- a/src/apps/structure-info/model.ts
+++ b/src/apps/structure-info/model.ts
@@ -126,6 +126,17 @@ export function printSequence(model: Model) {
console.log();
}
+export function printModRes(model: Model) {
+ console.log('\nModified Residues\n=============');
+ const { residueIndices, keys, data } = model.properties.modifiedResidues;
+ const { label_comp_id } = model.atomicHierarchy.residues;
+ for (let i = 0; i < residueIndices.length; i++) {
+ const rI = residueIndices[i], k = keys[i];
+ console.log(`Idx: ${rI} ${data.parent_comp_id.value(k)} -> ${label_comp_id.value(rI)}`);
+ }
+ console.log();
+}
+
export function printRings(structure: Structure) {
console.log('\nRings\n=============');
for (const unit of structure.units) {
@@ -186,6 +197,7 @@ async function run(mmcif: mmCIF_Database) {
printUnits(structure);
printRings(structure);
printLinks(structure, false, true);
+ printModRes(models[0]);
//printSecStructure(models[0]);
}
diff --git a/src/mol-model/structure/model/formats/mmcif.ts b/src/mol-model/structure/model/formats/mmcif.ts
index a50d29311..7cbaedb7d 100644
--- a/src/mol-model/structure/model/formats/mmcif.ts
+++ b/src/mol-model/structure/model/formats/mmcif.ts
@@ -11,7 +11,7 @@ import { Vec3 } from 'mol-math/linear-algebra';
import UUID from 'mol-util/uuid';
import Format from '../format';
import Model from '../model';
-import { AtomicConformation, AtomicData, AtomicSegments, AtomsSchema, ChainsSchema, ResiduesSchema } from '../properties/atomic';
+import { AtomicConformation, AtomicData, AtomicSegments, AtomsSchema, ChainsSchema, ResiduesSchema, AtomicHierarchy } from '../properties/atomic';
import { Entities } from '../properties/common';
import { ModelSymmetry } from '../properties/symmetry';
import { getAtomicKeys } from '../properties/utils/atomic-keys';
@@ -23,6 +23,7 @@ import { getSequence } from './mmcif/sequence';
import mmCIF_Format = Format.mmCIF
import { Task } from 'mol-task';
import { getSecondaryStructureMmCif } from './mmcif/secondary-structure';
+import { ModifiedResidues } from './mmcif/modified-residues';
function findModelBounds({ data }: mmCIF_Format, startIndex: number) {
const num = data.atom_site.pdbx_PDB_model_num;
@@ -118,6 +119,15 @@ function isHierarchyDataEqual(a: AtomicData, b: AtomicData) {
&& Table.areEqual(a.atoms as Table<AtomsSchema>, b.atoms as Table<AtomsSchema>)
}
+function modResProvider(format: mmCIF_Format, hierarchy: AtomicHierarchy, entities: Entities) {
+ let modres: ModifiedResidues | undefined = void 0;
+ return () => {
+ if (modres) return modres;
+ modres = new ModifiedResidues(format.data.pdbx_struct_mod_residue, hierarchy, entities);
+ return modres;
+ }
+}
+
function createModel(format: mmCIF_Format, bounds: Interval, previous?: Model): Model {
const hierarchyOffsets = findHierarchyOffsets(format, bounds);
const hierarchyData = createHierarchyData(format, bounds, hierarchyOffsets);
@@ -146,6 +156,8 @@ function createModel(format: mmCIF_Format, bounds: Interval, previous?: Model):
? format.data.entry.id.value(0)
: format.data._name;
+ const modRes = modResProvider(format, atomicHierarchy, entities);
+
return {
id: UUID.create(),
label,
@@ -158,7 +170,8 @@ function createModel(format: mmCIF_Format, bounds: Interval, previous?: Model):
coarseHierarchy: coarse.hierarchy,
coarseConformation: coarse.conformation,
properties: {
- secondaryStructure: getSecondaryStructureMmCif(format.data, atomicHierarchy)
+ secondaryStructure: getSecondaryStructureMmCif(format.data, atomicHierarchy),
+ get modifiedResidues() { return modRes() }
},
symmetry: getSymmetry(format)
};
diff --git a/src/mol-model/structure/model/formats/mmcif/modified-residues.ts b/src/mol-model/structure/model/formats/mmcif/modified-residues.ts
new file mode 100644
index 000000000..64f182b2b
--- /dev/null
+++ b/src/mol-model/structure/model/formats/mmcif/modified-residues.ts
@@ -0,0 +1,65 @@
+/**
+ * Copyright (c) 2018 mol* contributors, licensed under MIT, See LICENSE file for more info.
+ *
+ * @author David Sehnal <david.sehnal@gmail.com>
+ */
+
+import { mmCIF_Database } from 'mol-io/reader/cif/schema/mmcif'
+import { AtomicHierarchy } from '../../properties/atomic';
+import { Entities } from '../../properties/common';
+
+class ModifiedResidues {
+ private residueIndexMap = new Map<number, number>();
+
+ /** All residue indices within the given model that are modified. */
+ readonly residueIndices: ReadonlyArray<number>;
+ /** Indexed same as residueIndex and is key to ModifiedResidues.data table */
+ readonly keys: ReadonlyArray<number>;
+
+ /** Index into the data table. -1 if the residue is not modified. */
+ getKey(residueIndex: number): number {
+ return this.residueIndexMap.has(residueIndex) ? this.residueIndexMap.get(residueIndex)! : -1;
+ }
+
+ constructor(public data: mmCIF_Database['pdbx_struct_mod_residue'], hierarchy: AtomicHierarchy, entities: Entities) {
+ if (data._rowCount === 0) {
+ this.residueIndices = [];
+ this.keys = [];
+ return;
+ }
+
+ const { PDB_ins_code, auth_seq_id, _rowCount } = data;
+
+ const asym_id = data.label_asym_id.isDefined ? data.label_asym_id : data.auth_asym_id;
+ const comp_id = data.label_comp_id.isDefined ? data.label_comp_id : data.auth_comp_id;
+
+ const entityIds = entities.data.id.toArray();
+
+ const residueIndices: number[] = [];
+ const keys: number[] = [];
+
+ for (let i = 0; i < _rowCount; i++) {
+ const aId = asym_id.value(i);
+ const eIdx = getEntityId(hierarchy, entityIds, aId);
+ if (eIdx < 0) continue;
+ const key = hierarchy.findResidueKey(entityIds[eIdx], aId, comp_id.value(i), auth_seq_id.value(i), PDB_ins_code.value(i));
+ if (key >= 0) {
+ this.residueIndexMap.set(key, i);
+ residueIndices[residueIndices.length] = key;
+ keys[keys.length] = i;
+ }
+ }
+
+ this.residueIndices = residueIndices;
+ this.keys = keys;
+ }
+}
+
+function getEntityId(hierarchy: AtomicHierarchy, ids: ArrayLike<string>, asym_id: string) {
+ for (let i = 0, _i = ids.length; i < _i; i++) {
+ if (hierarchy.findChainKey(ids[i], asym_id) >= 0) return i;
+ }
+ return -1;
+}
+
+export { ModifiedResidues }
\ No newline at end of file
diff --git a/src/mol-model/structure/model/model.ts b/src/mol-model/structure/model/model.ts
index a20eaf796..8f60531e6 100644
--- a/src/mol-model/structure/model/model.ts
+++ b/src/mol-model/structure/model/model.ts
@@ -15,6 +15,7 @@ import { SecondaryStructure } from './properties/seconday-structure';
//import from_gro from './formats/gro'
import from_mmCIF from './formats/mmcif'
+import { ModifiedResidues } from './formats/mmcif/modified-residues';
/**
* Interface to the "source data" of the molecule.
@@ -37,7 +38,11 @@ interface Model extends Readonly<{
atomicConformation: AtomicConformation,
/** Various parts of the code can "cache" custom properties here */
- properties: { readonly secondaryStructure: SecondaryStructure } & { [customName: string]: any },
+ properties: {
+ readonly secondaryStructure: SecondaryStructure,
+ readonly modifiedResidues: ModifiedResidues,
+ [customName: string]: any
+ },
coarseHierarchy: CoarseHierarchy,
coarseConformation: CoarseConformation
diff --git a/src/mol-model/structure/model/properties/atomic/hierarchy.ts b/src/mol-model/structure/model/properties/atomic/hierarchy.ts
index fe50e990b..4327d22d4 100644
--- a/src/mol-model/structure/model/properties/atomic/hierarchy.ts
+++ b/src/mol-model/structure/model/properties/atomic/hierarchy.ts
@@ -65,7 +65,9 @@ export interface AtomicKeys {
// also index to the Entities table.
entityKey: ArrayLike<number>,
- findChainKey(entityId: string, label_asym_id: string): number
+ findChainKey(entityId: string, label_asym_id: string): number,
+
+ /** Unique number for each of the residue. Also the index of the 1st occurence of this residue. */
findResidueKey(entityId: string, label_asym_id: string, label_comp_id: string, auth_seq_id: number, pdbx_PDB_ins_code: string): number
}
--
GitLab