Automating Self-Driving Laboratory Growth ASSAY Protocol to Efficiently Measure Cell Antibiotic Resistance

Antibiotics have saved millions of lives over the past century, but their efficacy is decreasing because bacterial strains, especially in hospital settings, are developing resistance to antibiotics faster than new antibiotics can be produced.[1] Finding new antibiotic treatments that inhibit bacterial growth requires testing multiple combinations of antibiotics, which becomes strenuous and time-consuming for human researchers. Through the combination of Artificial Intelligence (AI) and robotics in self-driving laboratories, these treatments can be discovered more efficiently. AI can more intelligently select which treatments to test while the robotic experimentation platform can perform the necessary wet lab validations on a faster and larger scale than human researchers. With self-driving laboratories, these treatments can be discovered more efficiently due to a higher number of experiments being run continuously and machine learning algorithms to predict combinations. This paper seeks to create a fully automated biological assay with automated data processing to determine a bacterial cell’s susceptibility to an antibiotic. The robotic experimentation platform for the experiment includes a liquid handling robot, optical density reader, microplate manipulator, peeler, incubator, and sealer. All robotic actions are initiated from a Python script that invokes a Workflow Execution Interface (WEI) server to run workflows. Each workflow serves as a protocol of automated actions that are executed when the WEI server sends a request to the associated robot node. The automated experiment, intended to generate dose-response curves after a 12 hour incubation, is tested with Escherichia coli (E. coli) and tetracycline antibiotic as a proof-of-concept, and could later be used to test other antibiotic resistant bacterial strains. The results indicate that E. coli is susceptible to tetracycline antibiotic and there is negligible contamination because of the robot actions. These results were consistent with previous literature on the interactions between E. coli and tetracycline, demonstrating the promise of self-driving laboratories as a high throughput experimentation service for Argonne researchers.

[1] (Ventola, The antibiotic resistance crisis: part 1: causes and threats 2015)