Computational models for prediction of m6A sites using deep learning.

Area of Science:

• Molecular Biology
• Bioinformatics
• Genomics

Background:

• RNA modifications are vital for RNA diversity and regulation.
• N6-Methyladenosine (m6A) is the most prevalent internal mRNA modification in eukaryotes.
• Accurate m6A site identification is crucial for understanding its functional roles.

Purpose of the Study:

• To summarize existing machine learning and deep learning methods for m6A site prediction.
• To validate and compare various deep learning approaches, including underutilized and pre-trained models.
• To analyze dataset features and interpret model predictions for enhanced understanding of m6A site recognition.

Main Methods:

• Comprehensive literature review of machine learning and deep learning techniques for m6A site identification.
• Validation of multiple deep learning models on a benchmark dataset.
• Analysis of dataset features and model prediction interpretability.

Main Results:

• Deep learning models demonstrate superior performance in m6A site prediction compared to traditional methods.
• Underutilized and pre-trained deep learning models show significant potential.
• Model interpretation provides insights into the features driving m6A site recognition.

Conclusions:

• Deep learning approaches are highly effective for accurate m6A site identification.
• These models offer a powerful tool for advancing research in RNA modifications.
• Further exploration of deep learning holds promise for uncovering complex RNA regulatory mechanisms.