FoldX

Home > PDBe-KB Annotations > FoldX

FoldX

This page provides a description of the annotations provided by this PDBe-KB Consortium Member. These annotations are available via public FTP, from the PDBe graph database, the PDBe graph API, and on the PDBe-KB aggregated views of proteins.

---

Basic Information

What is the FTP URL of the provided annotations (JSON format)?

ftp://ftp.ebi.ac.uk/pub/databases/pdbe-kb/annotations/FoldX/

What is the URL of the data resource/software?

http://foldxsuite.crg.eu/

Who is the owner of the data resource/software?

Center for Genomic Regulation

What are the annotations provided?

Structural models of mutations related to folding and disease. Raw structures on the PDB are mutated using FoldX BuildModel command and the produced model is deposited. FoldX mutation routine performs mutations at sidechain level (the backbone remains intact) by annealing hundreds of observed rotamers in the amino acid to be mutated and subsequently moving its neighbours until minimization.

How regularly are these annotations updated?

Annually

---

Description of the data

Is there a "raw_score" provided? How to interpret it? Does it have a value range? Does it have a unit?

The value provided is the resulting FoldX measured ΔΔG upon mutation. Mutation destabilizing the protein folding are expected to have values greater than 1Kcal/Mol, and respectively, mutations stabilizing the protein will have negative values. In the case of deposited structures representing biomolecular complexes, the same apply for mutations placed in the interface, but stabilizing (or destabilizing) the represented complex.

Is there a "confidence_score" provided? How to interpret it? Does it have a value range?

Not yet. Structures with low resolution (>3A) or regions of structures with high BFactor can potentially be not accurate enough to measure energy changes upon mutation at sidechain level.

Is there a "confidence_level" provided? How is it decided?

Benchmarking of FoldX versus experimentally measured energy changes provides a confidence range of (± 0.4 Kcal/mol).

Are there benchmarking datasets? Are they available publicly? If yes, what is the URL?

Paper that describes in detail the benchmarking and forcefield construction: https://www.sciencedirect.com/science/article/abs/pii/S0022283602004424

How are these annotations collected/generated?

Mutations already uploaded comes from the structural mapping of all the mutation include in the Uniprot's humsavar data set. Next mutations to be uploaded will be obtained from VariBench (Nair, Preethy Sasidharan, and Mauno Vihinen. "V ari B ench: a benchmark database for variations." Human mutation 34.1 (2013): 42-49.)