Assessing the binding properties of CASP14 targets and models

Area of Science:

• Structural bioinformatics
• Computational biology
• Protein modeling

Background:

• Assessing the accuracy of protein models is crucial for their application in biological research.
• The Critical Assessment of protein Structure Prediction (CASP) provides a benchmark for protein structure modeling.
• Understanding how well CASP models retain native protein properties, especially binding sites, is an open question.

Purpose of the Study:

• To evaluate the ability of CASP models to preserve native protein binding site characteristics.
• To determine if CASP models can be used for predicting binding sites and facilitating ligand docking.
• To assess the utility of CASP models in understanding protein-protein interactions.

Main Methods:

• Calculated binding fingerprints based on residue interactions with small molecule probes.
• Quantified fingerprint similarity using correlation coefficients between X-ray structures and models.
• Performed ligand docking and protein-protein interaction analyses using CASP models and X-ray structures.

Main Results:

• Fingerprint similarity weakly correlates with global accuracy measures (GDT_TS).
• A GDT_TS score above 80 is necessary but not sufficient for conserving surface binding properties.
• CASP14 models demonstrated quality comparable to X-ray crystallography for identifying binding/regulatory sites and protein complexes.
• Ligand docking success is sensitive to binding interface details and may necessitate simulations for conformational changes.

Conclusions:

• CASP models are increasingly reliable for identifying functional sites and protein complexes.
• While models show promise, accurate ligand docking may require advanced simulation techniques to capture conformational dynamics.