CMD

The code for the model described in the paper "Conditional molecular design with deep generative models" https://arxiv.org/abs/1805.00108

Modified by Gonzalo Colmenarejo

Components

• run.py : main script
• SSVAE.py - model architecture
• mySSVAE.py - modification of the previous to deal with multiparameter conditioning
• preprocessing.py - functions for preprocessing
• exp0 : default run of CMD with original dataset and analysis of diversity, correctness and novelty of output
• exp1 : analysis of output (cor, div, nov) after training the CMD with increasing subsets of original training dataset
• exp3 : modified CMD for natural products; analyis of cor, div, nov of output
• exp4 : analysis of output (cor, div, nov) after training with orthogonal set vs clustered set
• cix : folder with myfuncs.py, a module created for performing all cheminformatic analyses

Dependencies

• TensorFlow
• NumPy
• RDKit
• chemfp
• scikit-learn
• pandas
• matplotlib

Environments

Two Anaconda environments are used to run the code: tf35 to run the CMD, and cix to run the analysis Jupyter notebooks