TRENDY: gene regulatory network inference enhanced by transformer

Area of Science:

• Computational Biology
• Systems Biology
• Bioinformatics

Background:

• Gene regulatory networks (GRNs) are essential for cellular function.
• Existing GRN inference methods include mechanism-based, information-based, and deep learning approaches.
• Current deep learning methods often neglect underlying gene expression mechanisms.

Purpose of the Study:

• To introduce TRENDY, a novel GRN inference method.
• To enhance mechanism-based GRN inference using transformer models.
• To demonstrate the broad applicability of transformer-based approaches in GRN inference.

Main Methods:

• Developed TRENDY by integrating transformer models with the WENDY approach.
• Evaluated TRENDY on simulated and experimental gene expression datasets.
• Applied the transformer-based enhancement to three other GRN inference methods.

Main Results:

• TRENDY significantly outperforms existing GRN inference methods.
• The transformer-based approach demonstrated broad potential for enhancing various inference techniques.
• Superior performance was observed on both simulated and real-world biological data.

Conclusions:

• TRENDY represents a significant advancement in mechanism-based GRN inference.
• Transformer models can effectively improve the accuracy and scope of GRN inference.
• The developed approach offers a powerful tool for dissecting complex gene regulatory mechanisms.