Wondering how accurate #AlphaFold models are?

We compared #AlphaFold predictions with experimental data. Many parts are great, but even high-confidence #AlphaFold predictions have distortions, domain movements, backbone errors, and side chain errors.

https://www.biorxiv.org/content/10.1101/2022.11.21.517405v1

@terwilltom Thanks for posting this! Hope this will shape how we can best use the predictions from #AlphaFold. There is great information in those predictions, but definite need for caution. I like that the #PDB is including that info, but with clearly marked distinction btw experimental and computational. Thanks!
#StructuralBiology #AlphaFold
@terwilltom many of those minute discrepancies are detrimental to drug discovery. I agree #AlphaFold provides a starting point for generating good theory, but that's pretty much what it can do now (we have been testing some of the models with an xxxFold-core and the mixed readout discourage us from spending more 💰 ⏱️ .)