SiteSeek: post-translational modification analysis using adaptive locality-effective kernel methods and new profiles.

Area of Science:

• Biochemistry
• Bioinformatics
• Computational Biology

Background:

• Post-translational modifications significantly impact protein structure and function.
• Protein phosphorylation is a common intracellular modification crucial for cellular signaling.
• Accurate prediction of phosphorylation sites is vital for biological research.

Purpose of the Study:

• To introduce SiteSeek, a novel machine learning-based predictor for protein phosphorylation sites.
• To utilize a unique compact evolutionary and hydrophobicity profile for enhanced prediction accuracy.
• To develop a more stable and accurate method compared to existing predictors.

Main Methods:

• Development of SiteSeek, a machine learning predictor.
• Training SiteSeek using a novel compact evolutionary and hydrophobicity profile.
• Performance evaluation against nine machine learning models and four established predictors using the PS-Benchmark_1 dataset.

Main Results:

• SiteSeek achieved 86.6% accuracy, 83.8% sensitivity, 92.5% specificity, and a 0.77 correlation coefficient.
• Outperformed existing predictors on the PS-Benchmark_1 dataset across key performance metrics.
• Demonstrated superior and stable predictive performance for major kinase families (CDK, CK2, PKA, PKC).

Conclusions:

• SiteSeek effectively identifies protein phosphorylation sites.
• The novel methods employed in SiteSeek offer significant improvements over existing predictors.
• SiteSeek's performance on the PS-Benchmark_1 dataset highlights its utility in biological research.