Hi Structural biologists!

Here is our article that helps you make the case that the experiment is still very much needed: https://www.nature.com/articles/s41592-023-02087-4

"AlphaFold predictions are valuable hypotheses and accelerate but do not replace experimental structure determination." Nature Methods (2023)

#Crystallography #MX #PDB #Alphafold #StructuralBiology @strucbio

Wondering just how accurate #AlphaFold models are?

For 102 recent structures, 80% of side chains agreed with the deposited model... but 20% did not and 7% were clearly incompatible with experimental data.

@buildmodels @PDBeurope @DeepMind

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

Wondering just how accurate #AlphaFold models are? Here is a handy chart!

The median error for high-confidence residues is just 0.6 Å!
Caution: 10% are off by over 2 Å.

#AlphaFold predictions are great #hypotheses

@buildmodels @PDBeurope @DeepMind

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

Wondering just how accurate #AlphaFold models are?

Here are high-confidence (pLDDT > 90) regions of #AlphaFold predictions compared with experimental data.

#AlphaFold predictions are great #hypotheses

@buildmodels @PDBeurope @DeepMind

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

#AlphaFold #AI with #Phenix #PredictAndBuild makes #Crystallography easier: All you need is your #Xray data and a sequence file.

#AlphaFold2 generates a #hypothesis, #PredictAndBuild compares it to the data, improves it and feeds it back to #AlphaFold2 to create a better #hypothesis.

(Fully automatic. Requires current nightly build of #Phenix from https://phenix-online.org)

https://www.biorxiv.org/content/10.1101/2022.11.18.517112v1

RT @d_brueckner
2. Sync your Twitter & #Mastodon account 🔁

Many won't want to give up Twitter yet (or ever), but doing 2 platforms is challenging

There's a simple tool to cross-post: https://moa.party 🎉

Everything you post on #Mastodon will show on Twitter & vice-versa 🤩