AlphaFold Protein Structure Database FAQs

Question 1

What file formats can I download using this skill?

Accepted Answer

You can download protein structures in both mmCIF (PDBx) and PDB formats, as well as JSON files for per-residue confidence scores and Predicted Aligned Error (PAE) matrices.

Question 2

What are pLDDT and PAE metrics in AlphaFold?

Accepted Answer

pLDDT measures per-residue confidence on a scale of 0-100, while PAE (Predicted Aligned Error) indicates the confidence level in the relative positions and orientations of different protein domains.

Question 3

How do I access bulk data for entire species?

Accepted Answer

The skill provides workflows to download full proteomes using taxonomy IDs (e.g., 9606 for humans) via Google Cloud Storage using gsutil commands.

Question 4

Does this skill require a local AlphaFold installation?

Accepted Answer

No, this skill retrieves pre-computed predictions from the public AlphaFold Database repository maintained by DeepMind and EMBL-EBI, eliminating the need for local GPU-intensive folding.

Question 5

Can I search for proteins using common names?

Accepted Answer

Yes, the skill includes utility functions to query UniProt to map protein or gene names to the specific accession IDs required for structure retrieval.

AlphaFold Protein Structure Database

Key Features

Use Cases

AlphaFold Protein Structure Database

Key Features

Use Cases