Accelerating the Discovery of Abiotic Vesicles with AI-Guided Automated Experimentation

Area of Science:

• Origin of life studies
• Biophysics
• Materials science

Background:

• Protocells, considered precursors to life, are hypothesized to form from self-assembling amphiphiles.
• Understanding vesicle formation is crucial for abiogenesis research.

Purpose of the Study:

• To accelerate the discovery of compositions and principles governing amphiphile-based vesicle formation.
• To develop an automated, AI-guided experimental workflow for studying protocell assembly.

Main Methods:

• High-throughput experimentation using a liquid handling robot for testing seven amphiphiles.
• Automated image analysis of vesicle formation via Multitemplate Multiscale Template Matching (MMTM).
• Gaussian process surrogate modeling and active learning to guide experiments and reduce uncertainty.

Main Results:

• Identified specific amphiphile mixtures, primarily equal amounts of trimethyl decylammonium and decylsulfate, forming vesicles at submillimolar concentrations.
• Determined that over 20% glycerol monodecanoate addition prevented vesicle formation, even at high amphiphile concentrations.

Conclusions:

• Demonstrated the efficacy of combining laboratory automation and AI for efficient discovery in protocell research.
• Provided insights into the compositional requirements and limitations for abiotic vesicle formation.