CoPhosK

CoPhosK predicts kinase-substrate associations (KSAs) for phosphopeptide substrates identified by mass spectrometry (MS) by leveraging co-phosphorylation dynamics.

Key Features:

• KSA prediction from MS: Predicts kinase-substrate associations for phosphopeptide substrates detected by mass spectrometry (MS).
• Co-phosphorylation analysis: Leverages collective dynamic signatures of kinases' substrates via correlation analysis of phosphopeptide intensity data.
• Naïve Bayes scoring: Employs a Naïve Bayes framework that incorporates priors based on known kinase-substrate associations to score candidate KSAs.
• Annotation-independent inference: Infers KSAs for substrates lacking existing annotations by analyzing dynamic co-phosphorylation patterns.
• Integration with static information: Combines dynamic MS-derived evidence with static information (sequences, structures, and interactions), improving prediction performance by approximately 35% for sites with available static data.
• Expanded coverage: Provides reliable predictions for substrates without static annotations, effectively tripling the number of KSAs derivable from experimental MS data.
• Benchmarking datasets: Validated using publicly available MS datasets from breast, colon, and ovarian cancer models.

Scientific Applications:

• Kinase-substrate network inference: Mapping KSAs from phosphoproteomic MS data to expand kinase-substrate interaction landscapes.
• Signaling pathway characterization: Elucidating complex signaling pathways by increasing coverage of kinase-substrate relationships.
• Cancer phosphoproteomics: Application to breast, colon, and ovarian cancer MS datasets for discovery and benchmarking of KSAs.

Methodology:

Performs correlation analysis of phosphopeptide MS intensity profiles to derive co-phosphorylation patterns and applies a Naïve Bayes classifier incorporating priors from known kinase-substrate associations to predict KSAs.