Synergistic enhancement of microalgal fatty acids via Batch Bayesian Optimization-driven combinatorial additive...

Area of Science:

• Biotechnology
• Microbial Production
• Lipid Synthesis

Background:

• Industrial fatty acid production by Schizochytrium sp. is limited by low fatty acid content.
• Optimizing chemical additives for lipid synthesis is challenging due to complex interactions and large experimental spaces.

Purpose of the Study:

• To develop an efficient machine learning-driven platform for optimizing chemical additive combinations.
• To enhance lipid synthesis and total fatty acid (TFA) content in Schizochytrium sp.

Main Methods:

• Utilized Batch Bayesian Optimization, a machine learning approach, to navigate high-dimensional parameter spaces.
• Iteratively optimized combinations of 13 candidate additives over six rounds.
• Validated optimized regimens through experimental testing.

Main Results:

• Achieved a 77.50% success rate in recommending effective additive combinations.
• Generated seven optimized regimens (binary, ternary, quaternary).
• Enhanced TFA content by approximately 50%, exceeding 50% of dry cell weight, a record for Schizochytrium sp. under shake-flask conditions, without inhibiting growth.

Conclusions:

• Machine learning-driven optimization effectively unlocks synergistic potential of chemical additives.
• This approach provides an efficient and scalable strategy for sustainable microbial lipid production.
• The study sets a new benchmark for TFA content in Schizochytrium sp. production.