iMul-kSite

iMul-kSite predicts multiple lysine post-translational modifications in peptide sequences to identify multi-label PTM sites for downstream proteomics and functional studies.

Key Features:

• Multi-PTM Prediction Capability: Predicts acetylation, crotonylation, methylation, succinylation, and glutarylation at lysine residues from peptide sequences.
• Data Optimization Techniques: Removes redundant majority-class samples via instance hardness analysis and refines feature representation using ANOVA F-Test combined with incremental feature selection.
• High Predictive Accuracy: Achieves 92.83% accuracy using the top 100 features, with an aiming rate of 93.36% and a coverage rate of 96.23%.

Scientific Applications:

• K-PTM Analysis: Supports investigation of lysine post-translational modifications (K-PTMs) in proteomics studies.
• Functional and Disease Studies: Aids exploration of the roles of lysine PTMs in cellular functions and disease mechanisms.
• Experimental Design and Drug Discovery: Provides multi-PTM site predictions to inform experimental designs and contribute evidence for drug discovery efforts.

Methodology:

Analyzes sequence-coupling information, applies instance hardness analysis to remove redundant majority-class samples, and optimizes features using ANOVA F-Test and incremental feature selection, employing the top 100 features for multi-label prediction.