Context-Aware Biosensor Design Through Biology-Guided Machine Learning and Dynamical Modeling.

Area of Science:

• Synthetic biology
• Metabolic engineering
• Biotechnology

Background:

• Achieving a global circular bioeconomy necessitates efficient, scalable bio-based processes for chemical production.
• Whole-cell biosensors, utilizing genetic circuits, are key to controlling cellular behavior and developing efficient cell factories.
• Existing models for biosensors like naringenin require further data for predicting dynamic responses in diverse applications.

Purpose of the Study:

• To engineer and characterize a library of FdeR biosensors for improved performance in bio-based production.
• To develop a mechanistic and machine learning-based predictive model for biosensor dynamic behavior.
• To optimize biosensor design and application conditions using a Design-Build-Test-Learn pipeline.

Main Methods:

• Assembled a library of FdeR biosensors and characterized their performance under varied conditions.
• Developed a mechanistic model for biosensor dynamic behavior under reference conditions.
• Implemented a machine learning model to predict context-dependent dynamic parameters.

Main Results:

• The Design-Build-Test-Learn pipeline successfully identified optimal condition combinations for desired biosensor specifications.
• The developed models enable prediction of biosensor performance across different regulatory elements, media, and supplements.
• Demonstrated the utility of biosensors for automated screening and dynamic regulation in engineered pathways.

Conclusions:

• This work enhances the understanding and application of whole-cell biosensors for precise measurement and regulation in bio-production.
• The developed DBTL pipeline and predictive models are crucial for optimizing biosensor design for circular bioeconomy applications.
• Engineered biosensors show significant potential for dynamic regulation of engineered production pathways for valuable molecules.