iRNA-m6A

iRNA-m6A predicts N6-methyladenosine (m6A) sites across multiple tissues in human, mouse, and rat to enable analysis of RNA methylation patterns.

Key Features:

• Ensemble Predictor: Employs an ensemble predictor leveraging high-throughput sequencing-derived data to identify m6A sites.
• Feature Encoding: Encodes RNA sequences using physical-chemical property matrix, mono-nucleotide binary encoding, and nucleotide chemical property.
• mRMR Feature Selection: Optimizes features using Minimum Redundancy Maximum Relevance (mRMR) to select relevant, non-redundant features.
• Machine Learning Model: Trains Support Vector Machine (SVM) models on the optimal feature subset and evaluates performance with 5-fold cross-validation.
• Species and Tissue Coverage: Covers prediction in human, mouse, and rat across tissues including brain, liver, and kidney.

Scientific Applications:

• Functional analysis of m6A: Identification of m6A sites to study roles in gene expression regulation and RNA stability.
• Tissue-specific epitranscriptomics: Characterization of tissue-specific m6A landscapes in brain, liver, and kidney.
• Comparative epitranscriptomics: Comparative analysis of m6A patterns across human, mouse, and rat.

Methodology:

Uses high-throughput sequencing-derived data with ensemble prediction; encodes sequences via physical-chemical property matrix, mono-nucleotide binary encoding, and nucleotide chemical property; applies mRMR for feature selection; trains and evaluates SVM models using 5-fold cross-validation.