skills.rest
google-deepmind
google-deepmindOFFICIAL
Explore all skills by this author →
WebsiteGitHub

alphafold-database-fetch-and-analyze

Fetch AlphaFold structures and PAE matrices for UniProt accessions and analyze pLDDT confidence and domain boundaries.

2.7k|283|Updated May 13, 2026
Search Tags:#pae#plddt#alphafold#uniprot#domain boundaries#structural confidence#protein disorder
One-click install
npx skills add https://github.com/google-deepmind/science-skills --skill alphafold-database-fetch-and-analyze
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: alphafold-database-fetch-and-analyze
Source: https://github.com/google-deepmind/science-skills/tree/main/skills/alphafold_database_fetch_and_analyze
Command: npx skills add https://github.com/google-deepmind/science-skills --skill alphafold-database-fetch-and-analyze

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires scienceskillscommon, and includes scripts (resource) components.

What problem does it solve?

This Skill solves the problem of quickly retrieving AlphaFold predicted structures for a specific protein and interpreting structural confidence and domain organization without manual, error-prone analysis.

Core Features & Use Cases

  • UniProt-ID driven retrieval: Downloads AlphaFold Database mmCIF structures and Predicted Aligned Error (PAE) matrices using a required UniProt accession, with guardrails to prevent misuse when only a name/sequence is provided.
  • pLDDT confidence analysis: Computes and reports global pLDDT plus fractions of very low, low, confident, and very high residues to assess overall ordering vs disorder.
  • PAE-based rigid domain boundaries: Detects and summarizes rigid global domains (including residue ranges) by applying a sliding-window PAE heuristic and merging sub-domains.
  • Operational edge-case warnings: Prominently relays canonical-vs-isoform substitutions and large-protein (>2700 aa) fragment warnings, and issues disorder-related warnings to discourage inappropriate downstream structural workflows.

Quick Start

Retrieve and analyze AlphaFold predictions for UniProt ID P00520 by instructing your agent: “Use alphafold-database-fetch-and-analyze to download the AlphaFold structure for UniProt P00520 and summarize pLDDT confidence and PAE domain boundaries.”

Frequently Asked Questions about alphafold-database-fetch-and-analyze

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I fetch an AlphaFold structure using a UniProt ID and analyze its pLDDT confidence?▼

To fetch an AlphaFold structure and analyze pLDDT confidence, provide a UniProt accession to retrieve the mmCIF file and PAE matrix. The analysis reports global pLDDT and residue fractions to indicate structural ordering versus disorder.

How does PAE matrix analysis detect rigid protein domain boundaries?▼

PAE matrix analysis detects rigid protein domain boundaries by applying a sliding-window heuristic to the Predicted Aligned Error data. It identifies and merges sub-domains, summarizing residue ranges for rigid global domains.

Can I retrieve AlphaFold predictions if I only have a protein name or sequence instead of a UniProt accession?▼

Retrieving AlphaFold predictions requires a UniProt accession, not just a protein name or sequence. Guardrails prevent misuse when only a name or sequence is provided, ensuring correct database queries for structural confidence reporting.

Does AlphaFold database fetching support large proteins or isoform variants?▼

AlphaFold database fetching supports large proteins and isoform variants but issues operational warnings. It prominently relays canonical-vs-isoform substitutions and large-protein fragment warnings for proteins over 2700 amino acids.

Why does AlphaFold pLDDT analysis issue disorder warnings for certain protein regions?▼

AlphaFold pLDDT analysis issues disorder warnings when very low confidence residues indicate significant disorder. These warnings discourage inappropriate downstream structural workflows by highlighting unreliable predicted regions.

What is the best way to report structural confidence for a protein using AlphaFold predictions?▼

The best way to report structural confidence is synthesizing pLDDT and PAE conclusions into a single summary. This approach combines global confidence metrics with rigid domain boundary detection for comprehensive structural assessment.

Related Skills

Explore More Education & Research Skills→
NeuroAIHubNeuroAIHub
269

alphafold-database-fetch-and-analyze

Fetches AlphaFold mmCIF structures and PAE matrices for UniProt IDs and analyzes them for confidence and domains.

Official
Advanced
HolobiomicsLabHolobiomicsLab
4

alphafold-database

Retrieve AlphaFold protein structures and confidence metrics by UniProt ID.

Official
Advanced
K-Dense-AIK-Dense-AI
33.0k

alphafold-database

Retrieve AlphaFold-predicted protein structures and confidence metrics by UniProt IDs.

Official
Intermediate
Yezez9Yezez9

alphafold-database

Retrieve AlphaFold protein structures and download PDB/mmCIF files by UniProt ID.

Community
Intermediate