m5UPred

m5UPred predicts 5-methyluridine (m^5U) sites in RNA sequences to enable identification of modification sites relevant to RNA regulation and disease studies.

Key Features:

• In Silico Prediction: Predicts m^5U sites from primary RNA sequences using computational models.
• Support Vector Machine (SVM) Algorithm: Employs an SVM classifier for site classification tasks.
• Biochemical Encoding Scheme: Integrates biochemical information into feature encoding for model input.
• Performance (AUC): Achieves area under the ROC curve (AUC) greater than 0.954 via 5-fold cross-validation and independent testing datasets.
• Cross-technique and Cross-cell-type Validation: Validated against miCLIP-Seq and FICC-Seq experimental datasets from HEK293 and HAP1 cell types, yielding average AUCs of 0.922 and 0.926 under mature mRNA mode.

Scientific Applications:

• Mapping m^5U modifications: Provides predicted m^5U positions to support identification and mapping of RNA modification sites for downstream experimental analysis.
• Studying biological roles and disease mechanisms: Supports investigation of m^5U involvement in processes such as stress responses and breast cancer by supplying candidate modification sites.

Methodology:

A Support Vector Machine classifier was trained on biochemical-encoded primary RNA sequence features and evaluated using 5-fold cross-validation and independent testing datasets.