Accelerating promoter identification and design by deep learning

Promoters are DNA sequences that govern the location, direction, and strength of gene transcription, playing a pivotal role in cellular growth and lifespan. Engineered promoters facilitate precise control of recombinant protein expression and metabolic pathway modulation for natural product biosynthesis. Traditional methods such as rational design and directed evolution have established the foundation for promoter engineering, and recent advances in deep learning (DL) have revolutionized the field. This review highlights the application of DL techniques for promoter identification, strength prediction, and de novo design using generative models. We describe how these tools are used and the impact of database quality, feature extraction, and model architecture on predictive accuracy. We discuss challenges and perspectives in developing robust models for promoter engineering.