ProCKSI

ProCKSI performs multi-dimensional comparison of protein structures by integrating multiple structural similarity measures to support structural genomics and comparative structural analysis.

Key Features:

• Integration of Multiple Similarity Measures: Incorporates Universal Similarity Metric (USM), Maximum Contact Map Overlap (MaxCMO), DaliLite, TM-align, Combinatorial Extension (CE), and FAST for comparative structural analysis.
• User-Defined Similarity Matrices: Accepts custom similarity matrices uploaded by users for inclusion in comparative analyses.
• Consensus Similarity Profile: Computes a consensus profile from multiple similarity measures to produce an aggregate view of structural similarity.
• Comprehensive Analysis Capabilities: Supports clustering, visualization, and detailed comparative analyses with validation against SCOP classifications.

Scientific Applications:

• Drug Design: Identifies structural similarities between proteins to inform target interactions and aid small-molecule design.
• Fold Prediction: Compares proteins to known structures to support fold prediction.
• Protein Clustering: Clusters proteins by structural similarity to support evolutionary studies and functional annotation.
• Evolutionary Studies: Analyzes structural similarities to infer evolutionary relationships among protein families.
• CASP Model Evaluation: Assesses similarity of predicted models to targets using a consensus approach for large structural deviations in CASP evaluations.
• Protein Kinase Classification Verification: Validates sequence-derived protein kinase classification schemes using structural similarity data.
• Dataset Analysis (RS126): Evaluates the impact of combining similarity measures on consensus assessment quality using datasets such as RS126.

Methodology:

Uses the Universal Similarity Metric (USM) based on algorithmic information theory; employs Contact Map Overlap (CMO) methods with integer linear programming via MaxCMO for optimal alignments and clustering validation against SCOP; integrates pairwise comparison and search methods DaliLite, TM-align, Combinatorial Extension (CE), and FAST; and computes consensus similarity profiles from the combined measures.