C-I-TASSER

C-I-TASSER predicts three-dimensional protein structures and functions by integrating deep-learning-derived contact maps into iterative threading and assembly refinement.

Key Features:

• Contact-map-guided prediction: Integrates deep-learning-based contact maps to refine protein structure predictions.
• Multiple Sequence Alignment (MSA) protocol: Employs a novel MSA generation protocol to construct deep sequence-profiles for accurate contact prediction, including for proteins lacking homologous templates.
• NeBcon meta-method: Incorporates NeBcon, a meta-method combining multiple contact predictors, including ResPRE which predicts contact maps by integrating precision-matrices with deep residual convolutional neural networks (CNNs).
• Optimized contact potential: Uses an optimized contact potential to guide structure assembly simulations.
• Iterative threading and assembly refinement: Performs iterative threading and assembly refinement guided by contact information to build structural models.

Scientific Applications:

• Free-modeling (FM) domains: Improves structure prediction for domains without close homologous templates, achieving average TM-scores 28% higher than I-TASSER for 50 FM domains in CASP13.
• Threading-based modeling (TBM) domains: Enhances threading modeling for domains with close homologous templates, producing TM-scores significantly higher than I-TASSER (P-value < 0.05).
• Limitations in quaternary and terminal regions: Faces challenges predicting quaternary structures of multi-domain proteins due to domain partitioning and reassembly, and shows reduced contact prediction accuracy in terminal regions when MSAs are sparse.

Methodology:

Uses deep-learning-based contact-map prediction (including ResPRE with precision-matrices and deep residual CNNs), the NeBcon meta-method combining multiple contact predictors, a novel MSA generation protocol for deep sequence-profiles, integration of contact maps into iterative threading and assembly refinement, and an optimized contact potential to guide assembly simulations.