ProQM-resample: improved model quality assessment for membrane proteins by limited conformational sampling

Area of Science:

• Computational biology
• Structural bioinformatics
• Protein modeling

Background:

• Model Quality Assessment Programs (MQAPs) are crucial for predicting protein structure quality.
• Existing methods include consensus-based and single-model approaches.
• Consensus methods improve accuracy but struggle to identify the best model and generate new structures.

Purpose of the Study:

• To develop a membrane protein-specific single-model MQAP.
• To improve protein model selection by enhancing side-chain packing.
• To create a method that can both sample and score new protein structures.

Main Methods:

• Introduced ProQM-resample, a single-model MQAP for membrane proteins.
• Coupled side-chain resampling with ProQM rescoring.
• Evaluated improvements in side-chain packing and model selection accuracy.

Main Results:

• Side-chain resampling improved packing in 96% of models.
• Model selection accuracy increased by 24% (Z-score sum from 25.0 to 31.1).
• ProQM-resample outperformed the consensus method Pcons.

Conclusions:

• ProQM-resample effectively improves side-chain packing and model selection for membrane proteins.
• Single-model methods offer advantages over consensus methods for structure generation and scoring.
• Improved side-chain quality is key to rescuing models with good backbones but poor side-chain packing.