AlphaFold Database

Protein structure prediction database by Google DeepMind / EMBL-EBI. Contains 200M+ predicted 3D structures with per-residue confidence scores (pLDDT).

What It Does

  • Provides predicted 3D protein structures for nearly every known protein
  • Per-residue confidence scores (pLDDT 0-100): >90 high confidence, 70-90 good, 50-70 low, <50 disordered
  • Predicted Aligned Error (PAE) for domain-level confidence
  • Downloadable PDB/mmCIF files

How to Use

Web Interface

  1. Go to https://alphafold.ebi.ac.uk
  2. Search by UniProt ID (e.g., Q7RTU9 for STRC) or gene name
  3. View 3D structure colored by pLDDT confidence
  4. Download structure files

API (Programmatic)

# Get structure by UniProt ID
curl https://alphafold.ebi.ac.uk/api/prediction/Q7RTU9
 
# Download PDB file
curl -O https://alphafold.ebi.ac.uk/files/AF-Q7RTU9-F1-model_v4.pdb

Python

import requests
r = requests.get("https://alphafold.ebi.ac.uk/api/prediction/Q7RTU9")
data = r.json()[0]
print(f"pLDDT: {data['globalMetricValue']}")
print(f"PDB URL: {data['pdbUrl']}")

Verified Status

VERIFIED — STRC (Q7RTU9) structure available (model v6). pLDDT at E1659: 95.69/100 — very high confidence, well-structured region. pLDDT profile shows high confidence in core domain (residues ~700-1775), low confidence in N-terminal disordered region.

STRC Research Usage

Results (April 2026)

  • PAE analysis DONE: E1659 has avg PAE 16.8Å (moderate), but 362 residues with PAE <5Å (confident interactions). 99 confident interactions within ±50 residues = E1659 is in a well-connected structural neighborhood.
  • ESMFold comparison DONE: cross-validated in STRC ESMFold Disorder Validation
  • Next: batch download ortholog structures, molecular dynamics

Connections