Awesome
Visual Neural Graph Fingerprints in Theano
<img src="https://github.com/debbiemarkslab/neural-fingerprint-theano/blob/master/data/images/neural_fing_img_github.png" width="700">This software package implements visual convolutional neural graph fingerprints in Theano/Lasagne.
The papers that describe the improvements can be found at:
Visual Neural Graph Fingerprints in Theano
The original publication is located at:
Convolutional Networks on Graphs for Learning Molecular Fingerprints
Requirements
Prerequisites include:
- Numpy, Scipy, Matplotlib: These are all found in Anaconda
- RDkit
- [Theano] (http://deeplearning.net/software/theano/install.html)
- Lasagne
Please see inside the code page for specifics of usage.
Examples
This code includes regression examples for visual convolutional neural fingerprints, convolutional neural fingerprints, LSTM-RNN on SMILES, and neural nets on standard fingerprints as well as the code to generate the visualizations.
Authors
The Theano/Lasagne implementation was written by Adam Riesselman. Please email me if you have any questions or concerns with this version.
The original neural graph fingerprint code was written by David Duvenaud, Dougal Maclaurin, and Ryan P. Adams; it can be found at their Github.