Computational Molecular Biology 2025, Vol.15, No.4, 193-207 http://bioscipublisher.com/index.php/cmb 204 combination" is detected will T cells release cytotoxic molecules. The result was that in mice, these cells precisely cleared tumors but hardly touched normal tissues. Such examples show for the first time that genetic circuits can become "programmed drugs". Figure 2 Complex dynamics of a forced oscillator are transformed to intricate spatial patterns (Adopted from Park et al., 2024) However, putting such "intelligent systems" into living organisms or natural environments is no easy task. Safety is always a prerequisite. Engineered bacteria may mutate, escape, or have unexpected interactions with the host system. Therefore, many designs have added suicide modules or external control switches, allowing the system to "shut down" at any time in case of any problem. The interference of complex physiological environments is more difficult to predict, which is why mathematical models and multiple rounds of experimental verification are so important. They can tell us in advance whether the circuit will still work as expected under different conditions. For future "intelligent cells" to truly enter clinical practice, they must first learn to maintain stability in complex environments. Looking ahead to the next stage, the status of models will only become more important, especially when they are combined with AI, the design approach may undergo a complete transformation. In the past, building a synthetic loop often relied on repeated trial and error through experiments. Now, model simulation can help us filter out a large number of inappropriate schemes, leaving only the most promising ones for experimentation. This is equivalent to moving half of the "design - build - test - learn (DBTL)" cycle into the computer. For instance, through the model, the response curves under different promoters or RBS intensities can be predicted in advance, thereby selecting a combination that is both fast and stable. This method significantly reduces the number of experiments. For complex circuits, it is a scenario where AI can truly shine. Machine learning can identify
RkJQdWJsaXNoZXIy MjQ4ODYzNA==