The data contained within this repository is the supporting information for Jack Simpson's PhD thesis "Computational Prediction of Gel Properties".

It is split by chapter in the following ways:

Chapter 2 - Classifcation Models

For each descriptor set explored in the chapter the data in this folder is as follows:

- Contains Pipeline Protocol protocols for descriptor calculation 
- Contains R scripts for data cleaning and model building (True and randomised models)
- Contains R objects containing the trained classifier models for each algorithm explored in Chapter 2. (True and randomised models)
- Contains R and Python scripts that carry out the SHAP model interpretation (all except fp_as_bits). 

Chapter 3 - Non-Bayesian Regression

- Contains Pipeline Protocol protocols for descriptor calculation 
- Contains R scripts for data cleaning and model building (True and randomised models)
- Contains R objects containing the trained regresion models for each algorithm explored in Chapter 3 and each data set generated. (True and randomised models)
- Contains R and Python scripts that carry out the SHAP model interpretation.

Chapter 4 - Bayesian Regression

- Contains Python scripts for:
	- Descriptor calculation
	- Model Building (True and randomised models)
	- SHAP analysis.

Chapter 5 - DeNovo Generation

- Contains python scripts for the genetic algorithm
- Contains the necessary scripts to train a Recurrent Neural Network using the approach undertaken in Chapter 5 and also contains a range of trained models. 

- Contanins benchmarking work including:
	- Scripts used to run the benchmarking
	- Outputs for each run
	- Scripts used to analyse the benchmarks.