iRice-MS

iRice-MS identifies post-translational modification (PTM) sites in the rice proteome to enable proteome-scale characterization of PTM-mediated regulation.

Key Features:

• Integrated XGBoost model: Employs eXtreme Gradient Boosting (XGBoost) as the predictive algorithm.
• Supported PTM types: Detects 2-hydroxyisobutyrylation, crotonylation, malonylation, ubiquitination, succinylation, and acetylation.
• Eight feature encoding schemes: Utilizes eight distinct feature encodings including sequence-based, physicochemical property-based, and spatial mapping information-based features.
• PTM-specific models: Builds separate models tailored to each modification by identifying optimal feature sets for each PTM type.
• Validation strategies: Assesses model performance using 5-fold cross-validation and independent dataset tests.
• Performance metric: Evaluated primarily by Area Under the Curve (AUC) and reported to outperform existing tools in AUC.
• Rice specificity: Models are developed specifically for rice proteins.

Scientific Applications:

• Proteome-scale PTM mapping in rice: Prediction of site-specific 2-hydroxyisobutyrylation, crotonylation, malonylation, ubiquitination, succinylation, and acetylation across rice proteins.
• Study of PTM-mediated regulation: Supporting investigations into the regulatory mechanisms of PTMs in plant biology.

Methodology:

Constructs integrated models based on eXtreme Gradient Boosting (XGBoost) using eight distinct feature encoding schemes (including sequence-based, physicochemical property-based, and spatial mapping information-based features), selects optimal feature sets for each PTM-specific model, and validates performance via 5-fold cross-validation and independent dataset testing with AUC comparisons.