This directory contains a Julia package, data, and scripts to support the paper "Case-Guided Sequential Assay Planning In Drug Discovery". It can be used for sequential planning of experimental designs, where each experiment requires some cost, and also produces some quantifiable information gain to refine a posterior distribution on target feature(s) of interest. Given historical data to define an implicit simulator of transition dynamics, and target(s) of interest, IBMDPDesigns.jl can generate a Pareto front of efficient experimental designs giving the optimal tradeoff between cost and information gain.
To build the IBMDPDesigns package, from the root directory, open a Julia REPL and use ] to enter the Pkg REPL. Then enter activate . to activate the directory as a Julia package, run instantiate to install the dependencies, and build to build the package. Press backspace to go back to the standard Julia REPL, and enter using IBMDPDesigns to precompile and import the package into the current scope. For more information on building Julia packages, please see the official Julia documentation.
After the package is built, see the README files in specific folders ADME_example, CNS_example, and benchmark for how to run the specific examples and reproduce results from the paper.