iRNA5hmC

iRNA5hmC predicts RNA 5-hydroxymethylcytosine (5hmC) sites from primary RNA sequences to enable computational characterization of 5hmC distributions in transcriptomes.

Key Features:

• Machine Learning Framework: Employs a support vector machine (SVM) model trained on sequence-derived features.
• Sequence-based Feature Algorithm: Utilizes a novel sequence-based feature algorithm to enable prediction from RNA primary sequences.
• Feature Representations: Implements two representations—k-mer spectrum to capture local nucleotide patterns and positional nucleotide binary vector to encode nucleotide positions.
• Feature Space Optimization: Applies a two-stage feature space optimization strategy to refine and enhance discriminative features.

Scientific Applications:

• 5hmC Site Prediction: Enables transcriptome-wide prediction of RNA 5-hydroxymethylcytosine (5hmC) sites.
• Complementary Analysis: Complements sequencing-based high-throughput experimental techniques for large-scale transcriptome analysis of 5hmC.
• Epigenetic and Functional Studies: Supports investigation of the biological roles and epigenetic regulation of RNA modifications.

Methodology:

Uses an SVM classifier trained with a novel sequence-based feature algorithm incorporating k-mer spectrum and positional nucleotide binary vector representations, together with a two-stage feature space optimization.