CPHmodels

CPHmodels predicts protein three-dimensional (3D) structures by single-template homology modeling to generate structural models for bioinformatics and structural biology analyses.

Key Features:

• Single-template homology modeling: Produces 3D models using a single-template homology-modeling strategy.
• Hybrid scoring functions: Combines the fast profile-profile scoring function from CPHmodels-2.0 with a novel remote homology-modeling algorithm.
• Two-tiered template identification: Attempts rapid modeling with the CPHmodels-2.0 scoring function and, if no suitable template is found, engages the remote homology-modeling algorithm.
• Benchmarked performance (CASP8): Generated models for 94% of CASP8 targets (117 out of 128).
• High-reliability model accuracy: 74% of models were classified as high reliability with an average RMSD of 4.6 Å to native structures.
• Low-reliability model accuracy: The remaining 26% of models had an average RMSD of 9.3 Å.
• Processing speed: Most queries are processed in under 20 minutes.

Scientific Applications:

• Drug discovery: Provides structural models to support target characterization and structure-based drug design.
• Enzyme design: Supplies 3D models for engineering and optimization of enzyme structure and function.
• Protein function characterization: Enables investigation of proteins with limited known structural information via homology-derived models.

Methodology:

Initial modeling uses a fast profile-profile scoring function from CPHmodels-2.0 for close homologs; if no suitable template is found the method applies a remote homology-modeling algorithm to identify templates and build single-template models.