Optimisation of surfactin yield in Bacillus using data-efficient active learning and high-throughput mass...

Area of Science:

• Synthetic Biology
• Metabolic Engineering
• Machine Learning

Background:

• The design-build-test-learn (DBTL) cycle is crucial for advancing synthetic biology.
• Optimizing production of complex molecules like surfactin requires sophisticated approaches.

Purpose of the Study:

• To apply active learning and metabolomics for optimizing surfactin production.
• To develop a data-driven framework for improving yields in complex biosynthetic pathways.

Main Methods:

• Implemented an active learning algorithm for media optimization within a DBTL framework.
• Utilized high-throughput metabolomics to analyze cellular biochemistry.
• Iteratively adjusted media composition based on algorithm-guided learning.

Main Results:

• Achieved a 160% increase in surfactin yield compared to baseline M9 media after three DBTL cycles.
• Metabolomics revealed biochemical insights into yield improvements and trade-offs.
• Identified positive associations between organic acids and surfactin production.

Conclusions:

• Active learning combined with metabolomics provides an effective data-driven strategy for optimizing biological product yields.
• The framework is suitable for complex pathways intractable to traditional methods.
• Understanding metabolic trade-offs is key for efficient bioproduction.