DRBP-EDP: classification of DNA-binding proteins and RNA-binding proteins using ESM-2 and dual-path neural network

Area of Science:

• Molecular Biology
• Bioinformatics
• Computational Biology

Background:

• Precise regulation of gene expression involves DNA-binding proteins (DBPs) and RNA-binding proteins (RBPs).
• Identifying these nucleic acid-binding proteins (NABPs) using traditional methods is challenging, time-consuming, and difficult to scale.
• Deep learning offers a more efficient approach to protein classification.

Purpose of the Study:

• To develop an efficient deep learning model for classifying NABPs, DBPs, and RBPs.
• To create high-quality datasets for protein classification tasks.
• To provide accessible tools for researchers in the field.

Main Methods:

• A phased classification method integrating ESM-2 with a dual-path neural network, named DRBP-EDP.
• Development of a refined dataset construction approach.
• Implementation of executable and web-based versions of the DRBP-EDP model.

Main Results:

• The DRBP-EDP model achieved 90.03% accuracy in classifying NABPs versus non-NABPs.
• The model demonstrated 89.56% accuracy in the second stage for classifying DBPs and RBPs.
• High-quality protein classification datasets were successfully created.

Conclusions:

• DRBP-EDP presents an efficient and accurate deep learning solution for classifying NABPs, DBPs, and RBPs.
• The developed datasets and accessible tools enhance usability for the scientific community.
• This approach overcomes limitations of traditional methods in identifying nucleic acid-binding proteins.