Research Outline

Synthetic genetic circuits with logic functions enable programmable control of gene expression, with applications spanning from bioindustry to therapeutics. DNA-writing circuits that use editing tools such as site-specific recombinases and CRISPR-based editors can implement logic-like DNA-state transitions, enabling multi-input and multistep programs with DNA-encoded memory.

Although simple DNA-writing circuits can be designed manually, the diversification of editing tools and functional genetic parts expands the design space and the range of implementable logic functions, making exhaustive exploration and heuristic tuning of genetic circuit designs increasingly impractical. Recent large language models (LLMs) may support such genetic circuit design by leveraging broad domain knowledge and reasoning that integrates multiple constraints. However, LLM-generated designs are not guaranteed to satisfy required DNA-state transitions or logical specifications. These challenges require computational verification of circuit designs before experimental implementation.

To address this, I am developing an autonomous genetic circuit design platform that couples LLM-based genetic circuit generation with computational verification (Fig. 1).