AI-based prediction of drug-gene interactions modulating tight junction integrity: A deep learning framework...

Area of Science:

• Biotechnology
• Bioinformatics
• Computational Biology

Background:

• Tight junctions are vital for epithelial and endothelial barrier integrity.
• Dysfunction of tight junctions is implicated in diseases like inflammatory bowel disease, asthma, and cancer.
• Identifying drugs that modulate tight junction function is crucial for therapeutic development.

Purpose of the Study:

• To develop a deep learning framework for predicting drug-induced modulation of tight junction integrity.
• To utilize multi-omics data for accurate prediction of drug-gene interactions affecting tight junctions.
• To identify novel drug candidates targeting tight junctions for disease treatment.

Main Methods:

• Preprocessed transcriptomic data from NCBI GEO to identify differentially expressed genes (DEGs).
• Employed network analysis to extract key hub genes.
• Trained a feedforward neural network model and evaluated its performance using metrics like AUC, CA, and F1-score.

Main Results:

• The neural network model achieved high performance with an AUC of 0.947, CA of 0.980, and F1-score of 0.969.
• Identified Cimifugin as a potential modulator of CLDN1, with Baicalein and Berberine also noted.
• The model demonstrated superior predictive power compared to traditional methods.

Conclusions:

• The developed deep learning framework offers a scalable, data-driven approach for predicting drug effects on tight junctions.
• This AI-based method facilitates drug discovery and personalized medicine for diseases involving tight junction disruption.
• The study highlights potential therapeutic strategies for conditions characterized by impaired barrier function.