iRNA-ac4C

iRNA-ac4C predicts N4-acetylcytidine (ac4C) modification sites in human mRNA to enable identification of ac4C-associated regulation of gene expression.

Key Features:

• Feature Extraction Methods: Extracts nucleotide composition, nucleotide chemical properties, and accumulated nucleotide frequency as input features for prediction.
• Optimal Feature Selection: Applies Minimum-Redundancy-Maximum-Relevance (mRMR) combined with incremental feature selection to determine the most informative feature subset.
• Classification Model: Trains a gradient boosting decision tree on the selected features and assesses performance using 10-fold cross-validation and independent testing sets.

Scientific Applications:

• Genome-wide identification: Enables identification of ac4C sites across human mRNA sequences.
• Regulatory mechanism analysis: Supports investigation of ac4C-mediated regulation of gene expression.
• Disease and functional studies: Facilitates exploration of roles of ac4C modifications in biological processes and potential disease contexts.

Methodology:

Computational steps include extracting nucleotide composition, nucleotide chemical properties, and accumulated nucleotide frequency; selecting features via mRMR with incremental feature selection; training a gradient boosting decision tree; and validating with 10-fold cross-validation and independent testing sets.