This library implements a web based GUI to DeepChem. As of present, it can be used to predict docking of ligands to proteins using DeepChem models pretrained on PDBBind. The GUI heavily relies on the molecular visualization library NGL and the Chemoinformatics library Kekule.js.

• Requirements
• Installation
• Usage
• Contributing to DeepChem
• DeepChem Publications
• About Us

• DeepChem
• NGL
• Flask

Installation from source is the only currently supported format.

1. Install DeepChem using the instructions here.

2. Install Flask: In the Anaconda environment you created in the previous step to install DeepChem, execute-

   conda install flask

3. Clone the deepchem-gui github repo:

   git clone https://github.com/deepchem/deepchem-gui.git && cd deepchem-gui

4. Install deepchem-gui:

   python setup.py install

1. In the Anaconda environment with deepchem-gui installed, execute:

   deepchem-gui server

   This should open your default browser and launch the application at http://127.0.0.1:5000/. Use the -h flag to reveal a full list of options.

2. Docking: Select ligand files and protein files using the file selection tool in Dock tab. At present, the only supported format for ligand files is .sdf and for protein files is .pdb. Docking takes approx. 5 minutes, following which the predicted scores are tabulated in the browser for all possible ligand and protein pairs. Clicking on a row loads the structures of the corresponding ligand and protein. Update the molecular visualizations using the options in the right panel as specified in these instructions.

3. Molecule editing: Clicking on the "Molecule Editor" option opens a Kekule.js molecule editor. Instructions to use the editor are available here.

4. Reaction visualization: Visualize SMILES strings and Reaction SMARTS stored in a CSV format using the respective tabs under "Visualize".

Note: This repository is under active development so bugs and surprises are likely. Kindly raise an issue on GitHub if you run into problems and we will try and resolve it asap. Alternatively you can also contribute to the repository by following these guidelines. Client side issues can be inspected using the JavaScript console of the browser while server side errors will be displayed in the shell.

1. Computational Modeling of β-secretase 1 (BACE-1) Inhibitors using Ligand Based Approaches
2. Low Data Drug Discovery with One-shot Learning

DeepChem is a package by the Pande group at Stanford. DeepChem was originally created by Bharath Ramsundar, and has grown through the contributions of a number of undergraduate, graduate, and postdoctoral researchers working with the Pande lab.