Spaces:
Running on Zero
Running on Zero
| """Fetching 3D protein structures: experimental structures UniProt | |
| cross-references (RCSB PDB) when available, otherwise the public AlphaFold | |
| DB's predicted model.""" | |
| from __future__ import annotations | |
| from dataclasses import dataclass | |
| import requests | |
| UNIPROT_ENTRY_URL = "https://rest.uniprot.org/uniprotkb/{accession}.json" | |
| ALPHAFOLD_PREDICTION_URL = "https://alphafold.ebi.ac.uk/api/prediction/{accession}" | |
| RCSB_PDB_URL = "https://files.rcsb.org/download/{pdb_id}.pdb" | |
| class PdbCrossReference: | |
| pdb_id: str | |
| method: str | None | |
| resolution: str | None | |
| chains: str | None | |
| class AlphaFoldPrediction: | |
| pdb_url: str | |
| mean_plddt: float | None | |
| model_created_date: str | None | |
| def fetch_pdb_cross_references(accession: str, timeout: float = 10.0) -> list[PdbCrossReference]: | |
| """Experimental structures UniProt cross-references to RCSB PDB for | |
| `accession` (often empty — most proteins have no solved structure).""" | |
| url = UNIPROT_ENTRY_URL.format(accession=accession.strip()) | |
| response = requests.get(url, timeout=timeout) | |
| response.raise_for_status() | |
| entry = response.json() | |
| references = [] | |
| for xref in entry.get("uniProtKBCrossReferences", []): | |
| if xref.get("database") != "PDB": | |
| continue | |
| properties = {p["key"]: p["value"] for p in xref.get("properties", [])} | |
| references.append( | |
| PdbCrossReference( | |
| pdb_id=xref.get("id"), | |
| method=properties.get("Method"), | |
| resolution=properties.get("Resolution"), | |
| chains=properties.get("Chains"), | |
| ) | |
| ) | |
| return references | |
| def fetch_alphafold_prediction(accession: str, timeout: float = 10.0) -> AlphaFoldPrediction | None: | |
| """Metadata for `accession`'s AlphaFold DB prediction (including the | |
| *current* model file URLs — AlphaFold DB periodically republishes | |
| models under a new version number, so this is fetched from the API | |
| rather than guessed/hardcoded), or None if it has no model.""" | |
| url = ALPHAFOLD_PREDICTION_URL.format(accession=accession.strip()) | |
| response = requests.get(url, timeout=timeout) | |
| if response.status_code == 404: | |
| return None | |
| response.raise_for_status() | |
| entries = response.json() | |
| if not entries: | |
| return None | |
| entry = entries[0] | |
| return AlphaFoldPrediction( | |
| pdb_url=entry["pdbUrl"], | |
| mean_plddt=entry.get("globalMetricValue"), | |
| model_created_date=entry.get("modelCreatedDate"), | |
| ) | |
| def fetch_structure_text(url: str, timeout: float = 20.0) -> str: | |
| """Raw PDB-format text from any public URL (AlphaFold DB or RCSB).""" | |
| response = requests.get(url, timeout=timeout) | |
| response.raise_for_status() | |
| return response.text | |
| def fetch_pdb_structure(pdb_id: str, timeout: float = 20.0) -> str: | |
| """An experimental structure from RCSB PDB, as PDB text.""" | |
| return fetch_structure_text(RCSB_PDB_URL.format(pdb_id=pdb_id.strip()), timeout=timeout) | |