Adapting machine-learning algorithms to design gene circuits.

Area of Science:

• Systems Biology
• Synthetic Biology
• Computational Biology

Background:

• Gene circuits perform diverse biological functions like oscillation, bistability, and signal response.
• Traditionally, researchers analyze existing circuits; this study reverses the approach to design circuits for specific functions.
• Discovering gene circuits computationally is challenging due to model complexity and numerous parameters.

Purpose of the Study:

• To develop a computational method for designing gene circuits based on desired functions.
• To accelerate the process of gene circuit discovery using machine learning algorithms.

Main Methods:

• Adapted machine learning algorithms, specifically gradient-descent optimization, to screen and design gene circuits.
• Developed a computational pipeline for rapid gene circuit discovery.

Main Results:

• Successfully designed gene circuits capable of executing a range of functions.
• Demonstrated the ability to recapitulate in vivo phenomena, such as oscillators.
• Enabled the design of circuits for complex synthetic biology tasks, including counting noisy biological events.

Conclusions:

• The computational pipeline facilitates systematic study of natural gene circuits.
• Enables automatic design of gene circuits for synthetic biology applications.
• The open-source Python module, GeneNet, is applicable to biological networks of any size and type.