Merge pull request #817 from molstar/pdb-assemblies-fix

Fix Archive PDB assembly loading

CHANGELOG.md

@@ -20,6 +20,7 @@ Note that since we don't clearly distinguish between a public and private interf
 - Parse HEADER record when reading PDB file
 - Support `ignoreHydrogens` in interactions representation
 - Add hydroxyproline (HYP) commonly present in collagen molecules to the list of amino acids
+- Fix assemblies for Archive PDB files (do not generate unique `label_asym_id` if `REMARK 350` is present)
 
 ## [v3.34.0] - 2023-04-16
 

src/mol-model-formats/structure/pdb/atom-site.ts

 /**
- * Copyright (c) 2019-2022 mol* contributors, licensed under MIT, See LICENSE file for more info.
+ * Copyright (c) 2019-2023 mol* contributors, licensed under MIT, See LICENSE file for more info.
  *
  * @author David Sehnal <david.sehnal@gmail.com>
  * @author Alexander Rose <alexander.rose@weirdbyte.de>
@@ -39,7 +39,7 @@ export function getAtomSiteTemplate(data: string, count: number) {
     };
 }
 
-export function getAtomSite(sites: AtomSiteTemplate, terIndices: Set<number>): { [K in keyof mmCIF_Schema['atom_site'] | 'partial_charge']?: CifField } {
+export function getAtomSite(sites: AtomSiteTemplate, terIndices: Set<number>, options: { hasAssemblies: boolean }): { [K in keyof mmCIF_Schema['atom_site'] | 'partial_charge']?: CifField } {
     const pdbx_PDB_model_num = CifField.ofStrings(sites.pdbx_PDB_model_num);
     const auth_asym_id = CifField.ofTokens(sites.auth_asym_id);
     const auth_seq_id = CifField.ofTokens(sites.auth_seq_id);
@@ -87,7 +87,9 @@ export function getAtomSite(sites: AtomSiteTemplate, terIndices: Set<number>): {
         if (asymIdCounts.has(asymId)) {
             // only change the chains name if there are TER records
             // otherwise assume repeated chain name use is from interleaved chains
-            if (terIndices.has(i)) {
+            // also don't change the chains name if there are assemblies
+            // as those require the original chain name
+            if (terIndices.has(i) && !options.hasAssemblies) {
                 const asymIdCount = asymIdCounts.get(asymId)! + 1;
                 asymIdCounts.set(asymId, asymIdCount);
                 currLabelAsymId = `${asymId}_${asymIdCount}`;

src/mol-model-formats/structure/pdb/to-cif.ts

 /**
- * Copyright (c) 2019-2021 mol* contributors, licensed under MIT, See LICENSE file for more info.
+ * Copyright (c) 2019-2023 mol* contributors, licensed under MIT, See LICENSE file for more info.
  *
  * @author David Sehnal <david.sehnal@gmail.com>
  * @author Alexander Rose <alexander.rose@weirdbyte.de>
@@ -54,6 +54,7 @@ export async function pdbToMmCif(pdb: PdbFile): Promise<CifFrame> {
 
     let modelNum = 0, modelStr = '';
     let conectRange: [number, number] | undefined = undefined;
+    let hasAssemblies = false;
     const terIndices = new Set<number>();
 
     for (let i = 0, _i = lines.count; i < _i; i++) {
@@ -152,6 +153,7 @@ export async function pdbToMmCif(pdb: PdbFile): Promise<CifFrame> {
                     }
                     helperCategories.push(...parseRemark350(lines, i, j));
                     i = j - 1;
+                    hasAssemblies = true;
                 }
                 break;
             case 'S':
@@ -208,7 +210,7 @@ export async function pdbToMmCif(pdb: PdbFile): Promise<CifFrame> {
         atomSite.label_entity_id[i] = entityBuilder.getEntityId(compId, moleculeType, asymIds.value(i));
     }
 
-    const atom_site = getAtomSite(atomSite, terIndices);
+    const atom_site = getAtomSite(atomSite, terIndices, { hasAssemblies });
     if (!isPdbqt) delete atom_site.partial_charge;
 
     if (conectRange) {