RNA-Protein Binding Sites Prediction via Multi Scale Convolutional Gated Recurrent Unit Networks

Area of Science:

• Molecular Biology
• Bioinformatics
• Computational Biology

Background:

• RNA-protein binding is crucial for gene expression regulation.
• Existing prediction methods struggle with sequence dependencies and diverse motif lengths of translation factors (TFs).

Purpose of the Study:

• To develop an advanced predictive model for RNA-protein binding that addresses limitations of current methods.
• To improve the accuracy of predicting RNA-protein binding preferences.

Main Methods:

• A novel predictive model combining multi-scale convolutional layers and a bidirectional gated recurrent unit (GRU) layer was developed.
• Multi-scale convolutional layers capture motif features of varying lengths.
• Bidirectional GRU layers model dependencies within subsequences.

Main Results:

• The proposed model demonstrated superior performance compared to four state-of-the-art methods.
• Investigated the impact of model architecture, including convolution layers and kernel sizes, on performance.
• Confirmed the effectiveness of the bidirectional GRU layer in enhancing prediction accuracy.

Conclusions:

• The developed model offers a significant advancement in predicting RNA-protein binding.
• The combination of multi-scale convolutions and bidirectional GRU effectively captures complex binding characteristics.
• This approach provides a more robust tool for studying gene expression regulation.