Inferring kinase-phosphosite regulation from phosphoproteome-enriched cancer multi-omics datasets

Area of Science:

• Molecular Biology
• Cancer Research
• Bioinformatics

Background:

• Phosphorylation is vital for eukaryotic cell signaling and disease, but identifying responsible kinases for detected phosphosites remains a challenge.
• High-throughput technologies detect numerous phosphosites, yet the specific kinases regulating them are largely unknown, hindering a complete understanding of cellular regulation.

Purpose of the Study:

• To identify and characterize kinase-phosphosite (KPS) pairs by analyzing quantitative data across multiple tumor and normal tissue datasets.
• To develop and validate a computational approach for predicting kinase-specific phosphorylation sites using KPS correlations.

Main Methods:

• Collected and analyzed quantitative data (transcriptional, protein, phosphorylation levels) from 10,159 tumor and 15 adjacent normal tissue samples.
• Investigated KPS pair linkages using experimental evidence and prediction tools, assessing correlations between kinase and phosphosite levels.
• Employed an oversampling method with an XGBoost algorithm (SMOTE-XGBoost) to predict potential kinase-specific phosphorylation sites.

Main Results:

• Found significant correlations between kinase expression/phosphorylation and phosphosite levels, validating KPS interconnections.
• Observed notably higher Spearman's correlation coefficients for KPS pairs in tumor samples compared to normal tissues.
• Successfully developed and applied the SMOTE-XGBoost approach to predict kinase-specific phosphorylation sites, integrating findings into the eKPI database.

Conclusions:

• Quantitative correlation analysis is effective for inferring kinase-phosphosite interconnections.
• Tumor-specific regulatory interactions between kinases and phosphosites are more pronounced than in normal tissues.
• The developed SMOTE-XGBoost method and the eKPI database offer valuable resources for advancing research on kinase-phosphosite regulatory relationships.