From 652f6c651b70bf00ac9a630d3f41a2686509097b Mon Sep 17 00:00:00 2001
From: Alexander Rose <alexander.rose@weirdbyte.de>
Date: Sat, 12 Mar 2022 10:58:33 -0800
Subject: [PATCH] fix wrong element assignment
---
CHANGELOG.md | 1 +
src/mol-model-formats/structure/cif-core.ts | 2 +-
src/mol-model-formats/structure/gro.ts | 5 +++-
src/mol-model-formats/structure/mol2.ts | 2 +-
src/mol-model-formats/structure/psf.ts | 5 +++-
src/mol-model-formats/structure/util.ts | 33 +++++++++++----------
6 files changed, 28 insertions(+), 20 deletions(-)
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 8b6c16297..8df9ea307 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -11,6 +11,7 @@ Note that since we don't clearly distinguish between a public and private interf
- Add ``CustomImportControls`` to left panel
- Add Zenodo import extension (load structures, trajectories, volumes, and zip files)
- Fix loading of some compressed files within sessions
+- Fix wrong element assignment for atoms with Charmm ion names
## [v3.3.1] - 2022-02-27
diff --git a/src/mol-model-formats/structure/cif-core.ts b/src/mol-model-formats/structure/cif-core.ts
index fd87abc7c..486b8450f 100644
--- a/src/mol-model-formats/structure/cif-core.ts
+++ b/src/mol-model-formats/structure/cif-core.ts
@@ -100,7 +100,7 @@ async function getModels(db: CifCore_Database, format: CifCoreFormat, ctx: Runti
const element_symbol = new Array<string>(atomCount);
for (let i = 0; i < atomCount; ++i) {
// TODO can take as is if type_symbol not given?
- element_symbol[i] = guessElementSymbolString(label.value(i));
+ element_symbol[i] = guessElementSymbolString(label.value(i), '');
}
typeSymbol = Column.ofStringArray(element_symbol);
formalCharge = Column.Undefined(atomCount, Column.Schema.int);
diff --git a/src/mol-model-formats/structure/gro.ts b/src/mol-model-formats/structure/gro.ts
index 33f0efac5..2fb95634b 100644
--- a/src/mol-model-formats/structure/gro.ts
+++ b/src/mol-model-formats/structure/gro.ts
@@ -27,6 +27,7 @@ function getBasic(atoms: GroAtoms, modelNum: number): BasicData {
const asymIds = new Array<string>(atoms.count);
const seqIds = new Uint32Array(atoms.count);
const ids = new Uint32Array(atoms.count);
+ const typeSymbol = new Array<string>(atoms.count);
const entityBuilder = new EntityBuilder();
const componentBuilder = new ComponentBuilder(atoms.residueNumber, atoms.atomName);
@@ -66,6 +67,8 @@ function getBasic(atoms: GroAtoms, modelNum: number): BasicData {
asymIds[i] = currentAsymId;
seqIds[i] = currentSeqId;
ids[i] = i;
+
+ typeSymbol[i] = guessElementSymbolString(atoms.atomName.value(i), atoms.residueName.value(i));
}
const auth_asym_id = Column.ofStringArray(asymIds);
@@ -87,7 +90,7 @@ function getBasic(atoms: GroAtoms, modelNum: number): BasicData {
label_entity_id: Column.ofStringArray(entityIds),
occupancy: Column.ofConst(1, atoms.count, Column.Schema.float),
- type_symbol: Column.ofStringArray(Column.mapToArray(atoms.atomName, s => guessElementSymbolString(s))),
+ type_symbol: Column.ofStringArray(typeSymbol),
pdbx_PDB_model_num: Column.ofConst(modelNum, atoms.count, Column.Schema.int),
}, atoms.count);
diff --git a/src/mol-model-formats/structure/mol2.ts b/src/mol-model-formats/structure/mol2.ts
index b2bfe635b..ac8b4e75c 100644
--- a/src/mol-model-formats/structure/mol2.ts
+++ b/src/mol-model-formats/structure/mol2.ts
@@ -41,7 +41,7 @@ async function getModels(mol2: Mol2File, ctx: RuntimeContext) {
for (let i = 0; i < atoms.count; ++i) {
type_symbol[i] = hasAtomType
? atoms.atom_type.value(i).split('.')[0].toUpperCase()
- : guessElementSymbolString(atoms.atom_name.value(i));
+ : guessElementSymbolString(atoms.atom_name.value(i), atoms.subst_name.value(i));
}
const atom_site = Table.ofPartialColumns(BasicSchema.atom_site, {
diff --git a/src/mol-model-formats/structure/psf.ts b/src/mol-model-formats/structure/psf.ts
index 362cf9f22..50f1b454b 100644
--- a/src/mol-model-formats/structure/psf.ts
+++ b/src/mol-model-formats/structure/psf.ts
@@ -21,6 +21,7 @@ function getBasic(atoms: PsfFile['atoms']) {
const asymIds = new Array<string>(atoms.count);
const seqIds = new Uint32Array(atoms.count);
const ids = new Uint32Array(atoms.count);
+ const typeSymbol = new Array<string>(atoms.count);
const entityBuilder = new EntityBuilder();
const componentBuilder = new ComponentBuilder(atoms.residueId, atoms.atomName);
@@ -68,6 +69,8 @@ function getBasic(atoms: PsfFile['atoms']) {
asymIds[i] = currentAsymId;
seqIds[i] = currentSeqId;
ids[i] = i;
+
+ typeSymbol[i] = guessElementSymbolString(atoms.atomName.value(i), atoms.residueName.value(i));
}
const atom_site = Table.ofPartialColumns(BasicSchema.atom_site, {
@@ -84,7 +87,7 @@ function getBasic(atoms: PsfFile['atoms']) {
label_entity_id: Column.ofStringArray(entityIds),
occupancy: Column.ofConst(1, atoms.count, Column.Schema.float),
- type_symbol: Column.ofStringArray(Column.mapToArray(atoms.atomName, s => guessElementSymbolString(s))),
+ type_symbol: Column.ofStringArray(typeSymbol),
pdbx_PDB_model_num: Column.ofConst(1, atoms.count, Column.Schema.int),
}, atoms.count);
diff --git a/src/mol-model-formats/structure/util.ts b/src/mol-model-formats/structure/util.ts
index f75908b05..b0ebfd663 100644
--- a/src/mol-model-formats/structure/util.ts
+++ b/src/mol-model-formats/structure/util.ts
@@ -47,30 +47,31 @@ export function guessElementSymbolTokens(tokens: Tokens, str: string, start: num
}
const reTrimSpacesAndNumbers = /^[\s\d]+|[\s\d]+$/g;
-export function guessElementSymbolString(str: string) {
+export function guessElementSymbolString(atomId: string, compId: string) {
// trim spaces and numbers, convert to upper case
- str = str.replace(reTrimSpacesAndNumbers, '').toUpperCase();
- const l = str.length;
+ atomId = atomId.replace(reTrimSpacesAndNumbers, '').toUpperCase();
+ const l = atomId.length;
- if (l === 0) return str; // empty
- if (l === 1) return str; // one char
+ if (l === 0) return atomId; // empty
+ if (l === 1) return atomId; // one char
if (l === 2) { // two chars
- if (str === 'NA' || str === 'CL' || str === 'FE' || str === 'SI' ||
- str === 'BR' || str === 'AS'
- ) return str;
+ if (atomId === 'NA' || atomId === 'CL' || atomId === 'FE' || atomId === 'SI' ||
+ atomId === 'BR' || atomId === 'AS'
+ ) return atomId;
}
- if (l === 3) { // three chars
- if (str === 'SOD') return 'NA';
- if (str === 'POT') return 'K';
- if (str === 'CES') return 'CS';
- if (str === 'CAL') return 'CA';
- if (str === 'CLA') return 'CL';
+ if (l === 3 && compId === atomId) { // three chars
+ if (atomId === 'SOD') return 'NA';
+ if (atomId === 'POT') return 'K';
+ if (atomId === 'CES') return 'CS';
+ if (atomId === 'CAL') return 'CA';
+ if (atomId === 'CLA') return 'CL';
}
- const c = str[0];
+ const c = atomId[0];
if (c === 'C' || c === 'H' || c === 'N' || c === 'O' || c === 'P' || c === 'S') return c;
return ''; // no reasonable guess, return empty string
-}
\ No newline at end of file
+}
+
--
GitLab