molML/Nano_Particles_Active_Learning — reverse-engineered prompt

I want to build a Python research project that reproduces a nanoparticle design workflow using Bayesian active learning. It should use the provided experimental cycle data to learn which PLGA PEG nanoparticle formulations lead to higher uptake in breast cancer cells, then suggest the next best formulations to test from a screening library.

Please set up the environment, organize the data, models, experiments, results, and figure generation so I can rerun the full study from the command line. Include scripts to train and evaluate the machine learning models, compare predictions across cycles, save acquired samples and predictions, and recreate the main result figures. I’d like the project to be clear enough for a scientist to run without digging through the code.

Use the methods from the paper as closely as possible, including Bayesian modeling and XGBoost style baselines if appropriate. Add a short README with setup, run commands, and citation info. Look up current docs online if you need to.