Controlling gene expression using AI designed Cis-regulatory elements.

Area of Science:

• Genomics
• Synthetic Biology
• Computational Biology

Background:

• Cis-regulatory elements (CREs) are vital for gene expression, but traditional design methods are inefficient.
• Deep learning advances are enhancing CRE prediction and enabling synthetic CRE design.

Purpose of the Study:

• To review the impact of AI on CRE design and prediction.
• To discuss current challenges and future directions in AI-driven CRE engineering.

Main Methods:

• Review of deep learning models, including specialized design models and DNA foundation models.
• Analysis of AI integration with high-throughput experimental data.
• Examination of emerging AI technologies like multimodal modeling and reinforcement learning.

Main Results:

• AI improves prediction accuracy and offers strategies for designing synthetic CREs like promoters and enhancers.
• AI models show progress in generating complex regulatory architectures.
• Current AI methods face challenges including data scarcity, prediction-experiment gaps, and limited interpretability.

Conclusions:

• AI holds significant potential for systematic and targeted synthetic CRE design.
• Future directions include addressing data quality, enhancing model interpretability, and exploring multimodal and reinforcement learning approaches.
• Overcoming current limitations is key to fully realizing AI's capabilities in regulatory element engineering.