This repository contains code accompanying the following paper on the Make-It robotic flow chemistry platform developed by the Jensen Research Group at MIT:
Bayesian Optimization of Computer-Proposed Multistep Synthetic Routes on an Automated Robotic Flow Platform (preprint)
The code is organized into two main folders:
hardware_control
contains Python classes developed to interface with common lab equipment for chemical synthesis (e.g., pumps, valves, HPLC)dragonfly_bayesopt_demo
contains an example Jupyter notebook demonstrating how to use the Dragonfly Bayesian optimization algorithm (developed by Kandasamy et al., paper) for single- and multi-objective optimization and to visualize response surfaces
Additional documentation is provided in each folder.
Significant contributions to the hardware control code were made by authors of the previous report on this platform.
This work was funded by the DARPA Make-It and Accelerated Molecular Discovery programs under contracts ARO W911NF-16-2-0023 and HR00111920025, respectively.