CATHER

CATHER integrates sequential profiles with predicted contact maps to improve template-based protein structure prediction via contact-assisted threading informed by deep-learning-derived contact predictions.

Key Features:

• Contact-Assisted Threading: Combines threading algorithms with contact information to guide template alignment and model selection.
• Integration of Sequential Profiles and Predicted Contact Maps: Uses both sequence-derived profiles and predicted residue-residue contact maps as complementary inputs.
• Deep Learning-Based Contact Prediction: Derives contact maps from a deep learning algorithm to supply structural restraints.
• Iterative Algorithm: Employs iterative refinement that synergizes sequence profiles and predicted contacts to optimize predictions.
• Template-Based Modeling: Produces three-dimensional protein models through template selection and threading informed by contacts and profiles.

Scientific Applications:

• Protein Structure Prediction: Predicts three-dimensional protein structures from amino acid sequences using contact-assisted template-based modeling.
• Benchmark Performance: Demonstrated high performance in CASP12 and CASP13, ranking within the Top 10 among 39 server groups for free modeling targets in CASP13.

Methodology:

Integrates deep-learning-predicted contact maps with sequential profiles and applies iterative contact-assisted threading for template-based protein structure modeling.