pub_soft/molstar
3
0
Fork 0
mirror of https://github.com/molstar/molstar.git synced 2026-06-04 13:30:24 +08:00
Code Issues Packages Projects Releases Wiki Activity

tests for SEQRES

Browse Source
This commit is contained in:
Ryan DiRisio
2026-03-17 15:20:44 -04:00
parent 003c5f8fb7
commit eabe4d46bc
1 changed files with 150 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

152
src/mol-model-formats/structure/_spec/pdb.spec.ts
View File

@@ -1,11 +1,20 @@
/**
* Copyright (c) 2019 mol* contributors, licensed under MIT, See LICENSE file for more info.
* Copyright (c) 2019-2026 mol* contributors, licensed under MIT, See LICENSE file for more info.
*
* @author Alexander Rose <alexander.rose@weirdbyte.de>
* @author Ryan DiRisio <rjdiris@gmail.com>
*/
import { TokenBuilder } from '../../../mol-io/reader/common/text/tokenizer';
import { TokenBuilder, Tokenizer } from '../../../mol-io/reader/common/text/tokenizer';
import { guessElementSymbolTokens } from '../util';
import { pdbToMmCif } from '../pdb/to-cif';
import { PdbFile } from '../../../mol-io/reader/pdb/schema';
/** Helper: build a PdbFile from a raw PDB string. */
function makePdb(pdbText: string): PdbFile {
const lines = Tokenizer.readAllLines(pdbText);
return { lines, variant: 'pdb' };
}
const records = [
['ATOM 19 HD23 LEU A 1 151.940 143.340 155.670 0.00 0.00', 'H'],
@@ -23,4 +32,143 @@ describe('PDB to-cif', () => {
expect(data.substring(tokens.indices[0], tokens.indices[1])).toBe(element);
}
});
});
describe('PDB SEQRES-to-label_seq_id alignment', () => {
it('assigns label_seq_id using SEQRES alignment for a complete chain', async () => {
// SEQRES declares ALA GLY VAL; ATOM records observe ALA GLY VAL at auth_seq_id 1-3.
// All residues match, so label_seq_id should be 1, 2, 3.
const pdb = makePdb([
'SEQRES 1 A 3 ALA GLY VAL ',
'ATOM 1 CA ALA A 1 1.000 2.000 3.000 1.00 0.00 C ',
'ATOM 2 CA GLY A 2 4.000 5.000 6.000 1.00 0.00 C ',
'ATOM 3 CA VAL A 3 7.000 8.000 9.000 1.00 0.00 C ',
'END ',
].join('\n'));
const cif = await pdbToMmCif(pdb);
const atomSite = cif.categories['atom_site'];
const labelSeqId = atomSite.getField('label_seq_id')!;
expect(labelSeqId.int(0)).toBe(1);
expect(labelSeqId.int(1)).toBe(2);
expect(labelSeqId.int(2)).toBe(3);
});
it('offsets label_seq_id when leading SEQRES residues are missing from ATOM', async () => {
// SEQRES: MET ALA GLY VAL (4 residues)
// ATOM: only ALA GLY VAL at auth_seq_id 2-4.
// MET is unobserved, so ALA aligns to SEQRES position 2, GLY→3, VAL→4.
const pdb = makePdb([
'SEQRES 1 A 4 MET ALA GLY VAL ',
'ATOM 1 CA ALA A 2 1.000 2.000 3.000 1.00 0.00 C ',
'ATOM 2 CA GLY A 3 4.000 5.000 6.000 1.00 0.00 C ',
'ATOM 3 CA VAL A 4 7.000 8.000 9.000 1.00 0.00 C ',
'END ',
].join('\n'));
const cif = await pdbToMmCif(pdb);
const atomSite = cif.categories['atom_site'];
const labelSeqId = atomSite.getField('label_seq_id')!;
expect(labelSeqId.int(0)).toBe(2); // ALA -> SEQRES pos 2
expect(labelSeqId.int(1)).toBe(3); // GLY -> SEQRES pos 3
expect(labelSeqId.int(2)).toBe(4); // VAL -> SEQRES pos 4
});
it('handles an internal gap in observed residues', async () => {
// SEQRES: ALA GLY VAL LEU (4 residues)
// ATOM: ALA VAL LEU (GLY is missing in the middle)
// Alignment: ALA→1, VAL→3, LEU→4
const pdb = makePdb([
'SEQRES 1 A 4 ALA GLY VAL LEU ',
'ATOM 1 CA ALA A 1 1.000 2.000 3.000 1.00 0.00 C ',
'ATOM 2 CA VAL A 3 4.000 5.000 6.000 1.00 0.00 C ',
'ATOM 3 CA LEU A 4 7.000 8.000 9.000 1.00 0.00 C ',
'END ',
].join('\n'));
const cif = await pdbToMmCif(pdb);
const atomSite = cif.categories['atom_site'];
const labelSeqId = atomSite.getField('label_seq_id')!;
expect(labelSeqId.int(0)).toBe(1); // ALA -> SEQRES pos 1
expect(labelSeqId.int(1)).toBe(3); // VAL -> SEQRES pos 3
expect(labelSeqId.int(2)).toBe(4); // LEU -> SEQRES pos 4
});
it('assigns label_seq_id for multiple chains independently', async () => {
// Chain A SEQRES: MET ALA, ATOM: ALA (at pos 2)
// Chain B SEQRES: GLY VAL LEU, ATOM: GLY VAL LEU (at pos 1,2,3)
const pdb = makePdb([
'SEQRES 1 A 2 MET ALA ',
'SEQRES 1 B 3 GLY VAL LEU ',
'ATOM 1 CA ALA A 2 1.000 2.000 3.000 1.00 0.00 C ',
'TER 2 ALA A 2 ',
'ATOM 3 CA GLY B 1 4.000 5.000 6.000 1.00 0.00 C ',
'ATOM 4 CA VAL B 2 7.000 8.000 9.000 1.00 0.00 C ',
'ATOM 5 CA LEU B 3 10.000 11.000 12.000 1.00 0.00 C ',
'END ',
].join('\n'));
const cif = await pdbToMmCif(pdb);
const atomSite = cif.categories['atom_site'];
const labelSeqId = atomSite.getField('label_seq_id')!;
// Chain A: ALA aligns to SEQRES position 2
expect(labelSeqId.int(0)).toBe(2);
// Chain B: GLY→1, VAL→2, LEU→3
expect(labelSeqId.int(1)).toBe(1);
expect(labelSeqId.int(2)).toBe(2);
expect(labelSeqId.int(3)).toBe(3);
});
it('populates pdbx_unobs_or_zero_occ_residues for missing SEQRES residues', async () => {
// SEQRES: MET ALA GLY VAL (4 residues), ATOM: ALA GLY (pos 2,3)
// MET (pos 1) and VAL (pos 4) should appear in unobs
const pdb = makePdb([
'SEQRES 1 A 4 MET ALA GLY VAL ',
'ATOM 1 N ALA A 2 1.000 2.000 3.000 1.00 0.00 N ',
'ATOM 2 CA ALA A 2 1.500 2.500 3.500 1.00 0.00 C ',
'ATOM 3 C ALA A 2 2.000 3.000 4.000 1.00 0.00 C ',
'ATOM 4 N GLY A 3 4.000 5.000 6.000 1.00 0.00 N ',
'ATOM 5 CA GLY A 3 4.500 5.500 6.500 1.00 0.00 C ',
'ATOM 6 C GLY A 3 5.000 6.000 7.000 1.00 0.00 C ',
'END ',
].join('\n'));
const cif = await pdbToMmCif(pdb);
const unobs = cif.categories['pdbx_unobs_or_zero_occ_residues'];
expect(unobs).toBeDefined();
const unobsSeqId = unobs.getField('label_seq_id')!;
const unobsCompId = unobs.getField('label_comp_id')!;
expect(unobsSeqId.rowCount).toBe(2);
// Positions 1 (MET) and 4 (VAL) are unobserved
expect(unobsSeqId.int(0)).toBe(1);
expect(unobsCompId.str(0)).toBe('MET');
expect(unobsSeqId.int(1)).toBe(4);
expect(unobsCompId.str(1)).toBe('VAL');
});
it('handles multiple atoms per residue correctly', async () => {
// SEQRES: MET ALA, ATOM: ALA with N, CA, C atoms
// All 3 atoms of ALA should get label_seq_id = 2
const pdb = makePdb([
'SEQRES 1 A 2 MET ALA ',
'ATOM 1 N ALA A 2 1.000 2.000 3.000 1.00 0.00 N ',
'ATOM 2 CA ALA A 2 1.500 2.500 3.500 1.00 0.00 C ',
'ATOM 3 C ALA A 2 2.000 3.000 4.000 1.00 0.00 C ',
'END ',
].join('\n'));
const cif = await pdbToMmCif(pdb);
const atomSite = cif.categories['atom_site'];
const labelSeqId = atomSite.getField('label_seq_id')!;
expect(labelSeqId.int(0)).toBe(2);
expect(labelSeqId.int(1)).toBe(2);
expect(labelSeqId.int(2)).toBe(2);
});
});