Chemception

CNN model based on Google Inception model to predict regression values of regression values of Tetrahymena pyriformis IGC50 from 2D images of molcules

This post is inspired from the following papers and blog posts.

1. https://www.wildcardconsulting.dk/useful-information/learn-how-to-teach-your-computer-to-see-chemistry-free-chemception-models-with-rdkit-and-keras/

2. https://arxiv.org/abs/1706.06689

3. https://arxiv.org/abs/1710.02238

Description

Activity is predicted from the 2D molecular images and the R^2 value of 0.71 is obtained. The data set is about Tetrahymena pyriformis IGC50 48h toxicity.

System Setup

Following packages should be installed before running the code.

Install the following packges.
pip install jupyter
Pip install Tensorflow
Pip install Keras
Pip install sklearn
Pip install PIL
Pip install pandas
Pip install numpy
Pip install scipy
Pip install openpyxl
Pip install xlsxwriter
Pip install h5py
Pip install matplotlib
pip install rdkit