DNA-Diffusion: Leveraging Generative Models for Controlling Chromatin Accessibility and Gene Expression via Synthetic...

Area of Science:

• Genomics
• Synthetic Biology
• Bioinformatics

Background:

• Precise control of gene expression is crucial for genomics and synthetic biology.
• Generative AI offers new methods for designing synthetic DNA sequences to modulate gene expression.

Approach:

• Developed a cell type-specific diffusion model to generate synthetic 200 bp DNA regulatory elements.
• Generated sequences based on chromatin accessibility data across various cell types.
• Evaluated sequences using transcription factor binding site composition, predicted chromatin accessibility, and gene expression activation capacity.

Key Points:

• The DNA-Diffusion framework successfully generated DNA sequences with cell type-specific regulatory potential.
• Generated sequences mimic properties of endogenous regulatory elements.
• State-of-the-art prediction models confirmed the regulatory capacity of the designed sequences.

Conclusions:

• DNA-Diffusion enables robust generation of cell type-specific regulatory DNA sequences.
• This approach has significant potential for advancing mammalian synthetic biology.
• The method paves the way for novel precision gene therapy applications.