Hold out the genome: a roadmap to solving the cis-regulatory code.

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• Gene expression is controlled by transcription factors interacting with cis-regulatory DNA sequences.
• The 'cis-regulatory code' dictates gene expression timing, location, and level, presenting significant complexity.
• Advances in functional genomics and machine learning offer new tools for deciphering this code.

Purpose of the Study:

• To propose a roadmap for solving the cis-regulatory code.
• To highlight the utility of synthetic DNA in conjunction with machine learning for this purpose.

Main Methods:

• Utilizing machine learning models trained on both genomic and synthetic DNA sequences.
• Employing massively parallel assays with synthetic DNA for targeted querying.
• Comparing predictive performance of models trained on different data sources.

Main Results:

• Models trained on synthetic DNA can predict genomic activity effectively, sometimes outperforming genome-trained models.
• Synthetic DNA expands the accessible sequence space beyond genomic limitations.
• Designed synthetic sequences allow for targeted improvements to predictive models.

Conclusions:

• Solving the cis-regulatory code requires moving beyond solely genomic data.
• A combination of machine learning and high-throughput synthetic DNA assays is a promising strategy.
• This integrated approach can lead to a more complete understanding of gene regulation.