GPredictor

GPredictor predicts internal N7-methylguanosine (m7G) modifications in RNA sequences to identify post-transcriptional modification sites and support studies of RNA processing.

Key Features:

• Machine Learning Model: Employs a machine learning model to predict internal m7G sites with high accuracy.
• Feature Extraction Methods: Utilizes Pseudo dinucleotide composition, Pseudo k-tuple composition, K monomeric units, Ksnpf frequency, and Nucleotide chemical property for sequence representation.
• Optimization Algorithm: Uses Random Forest to identify the optimal subset of features for prediction.
• Predictive Performance: Implements an SVM-based predictor trained on the top 240 selected features and reports superior performance compared to iRNA-m7G.
• Validation Methods: Validated using 10-fold cross-validation, Jackknife test, and independent testing.

Scientific Applications:

• Mechanistic studies: Provides insights into the mechanisms and distribution of internal m7G sites in RNA.
• Experimental validation and medical research: Supports experimental validation and research on the biological significance and potential medical implications of m7G modifications.

Methodology:

Feature extraction using Pseudo dinucleotide composition, Pseudo k-tuple composition, K monomeric units, Ksnpf frequency, and Nucleotide chemical property; Random Forest for feature selection to identify the top 240 features; SVM trained on the selected 240 features; validation by 10-fold cross-validation, Jackknife test, and independent testing.