structure-analysis

Protein structure analysis

One PDB entry yields different residue counts, interface lists and hydrogen-bond totals depending on how waters, HETATM records, alternate locations and symmetry copies were treated. Those choices, and every distance or angle cutoff, are part of the result and are reported with it.

Files and identity

1. Prefer PDBx/mmCIF over the legacy PDB format: PDB files break above 99,999 atoms and 62 chains, truncate chain IDs to one character and have no entity concept. Download https://files.rcsb.org/download/<id>.cif or use Biopython PDBList().retrieve_pdb_file("1abc", pdir=".", file_format="mmCif"). Record the entry id, the model used (NMR ensembles have many; state the index) and the download date.
2. Chains and entities: mmCIF has label_asym_id (internal chain), auth_asym_id (the author chain that papers and viewers use) and label_entity_id; a homodimer is one entity with two chains. Residue numbering is auth_seq_id (author numbering with gaps, insertion codes and negatives) or label_seq_id (1..n over the entity sequence). UniProt numbering is a third scheme, mapped by SIFTS. Say which one every residue number in the report follows.
3. HETATM policy: ligands, ions, modified residues inside the polymer (MSE, SEP, TPO) and waters (HOH, DOD) are all HETATM. Decide and state whether waters are removed, ions kept, and modified residues counted as protein. Biopython encodes this in residue.id[0] (" " standard, "W" water, "H_XXX" hetero) and Bio.PDB.Polypeptide.is_aa(res, standard=True); gemmi offers st.remove_waters(), st.remove_ligands_and_waters(), st.remove_hydrogens() and Residue.het_flag. Call st.setup_entities() after reading PDB files with gemmi.
4. Alternate locations: pick one conformer consistently. Biopython's PDBParser wraps them in DisorderedAtom objects that default to the highest occupancy (atom.is_disordered()); gemmi st.remove_alternative_conformations() keeps the first; MDAnalysis selects with altloc A. Counting all altlocs doubles some contacts.
5. Insertion codes: a Biopython residue id is (hetflag, resseq, icode), so 52 and 52A are distinct residues; parsing the number alone merges them. Antibody numbering schemes rely on them.
6. Assemblies: the asymmetric unit may hold half a dimer or two copies of a monomer. The biological assembly applies the operators in _pdbx_struct_assembly (REMARK 350). gemmi: gemmi.make_assembly(st.assemblies[0], st[0], gemmi.HowToNameCopiedChain.AddNumber) or st.transform_to_assembly("1", how); PyMOL set assembly, 1 before fetch; ChimeraX sym #1 assembly 1; RCSB serves <id>-assembly1.cif. Interfaces across symmetry mates exist only in the assembly. Report which assembly, or that the asymmetric unit was used.
7. Hydrogens: crystal structures rarely include them. Angle-based hydrogen-bond criteria need added hydrogens (PDBFixer or OpenMM Modeller.addHydrogens, PyMOL h_add, ChimeraX addh); heavy-atom-only criteria are the alternative. State which.