Assessment of the CASP14 assembly predictions

Abstract

In CASP14, 39 research groups submitted more than 2500 3D models on 22 protein complexes. In general, the community performed well in predicting the fold of the assemblies (for 80% of the targets), although it faced significant challenges in reproducing the native contacts. This is especially the case for the complexes without whole-assembly templates. The leading predictor, BAKER-experimental, used a methodology combining classical techniques (template-based modeling, protein docking) with deep learning-based contact predictions and a fold-and-dock approach. The Venclovas team achieved the runner-up position with template-based modeling and docking. By analyzing the target interfaces, we showed that the complexes with depleted charged contacts or dominating hydrophobic interactions were the most challenging ones to predict. We also demonstrated that if AlphaFold2 predictions were at hand, the interface prediction challenge could be alleviated for most of the targets. All in all, it is evident that new approaches are needed for the accurate prediction of assemblies, which undoubtedly will expand on the significant improvements in the tertiary structure prediction field.

Prediction of protein assemblies, the next frontier: The CASP14‐CAPRI experiment

“…These evaluations are carried out using criteria and evaluation protocols agreed upon by the CAPRI community. A separate evaluation of the CASP14 assembly prediction performance, reported at the CASP14 meeting and in this Special Issue, 45 was performed by the CASP assembly assessment team in collaboration with the CASP prediction center. We wish to highlight the very fruitful collaboration that took place between the CASP teams and the CAPRI assessment in defining the prediction problem for complex targets, discussing evaluation strategies and comparing assessment results.…”

The Structure of the Arabidopsis PEX4-PEX22 Peroxin Complex—Insights Into Ubiquitination at the Peroxisomal Membrane

“…The use of modern modeling methods such as AlphaFold ( Jumper et al, 2021 ) is expected to help in future structure determinations. In fact, the PEX22 structure was submitted as part of the CASP14 (Critical Assessment of Structure Prediction) competition ( Ozden et al, 2021 ), and retrospective analysis showed that the best predicted models of PEX22 alone provided better initial molecular replacement solutions than did our PEX4 homology model.…”

Cryo‐EM targets in CASP14

“…Unfortunately, despite the high accuracy of the individual protein level predictions, the prediction of quaternary structures was not successful, possibly due to the size and complexity of those assemblies. 7 Refinement of the submitted CASP models in the experimental density shows that the models could be improved to the point of approaching the quality of the reference structures (and beyond in some structural elements), thus indicating that high-quality models from CASP predictors can be a good starting point for structure refinement. These structures often represent large complexes, where many proteins have to be predicted, some of which can only be modeled accurately in the context of other proteins.…”