Software

Please visit our GitHub to view all our code. Feel free to join any of the repositories you like, find issues, and suggest changes. We believe that an open community and open software is essential for science. Almost all of our code is OpenBSD and free for academic and educational purposes. For use in industry, please contact gchopra@purdue.edu. Thank you.

MINT

Molecular Interaction with New Technology 

MINT is a virtual reality chemistry gaming platform for visualizing and manipulating chemical structures. 

[Website]

AI/ML for Drug Design

Energy-based Graph Neural Network for Immunomodulators

Energy based Graph Neural Network – (based on GNN + docking). This example is for synthesizing molecules for cancer immunotherapy and we were able to do scaffold hopping.

[GitHub] -- New changes coming soon!

AI/ML for Chemical Reactivity

Interpretable Machine Learning for N-sulfonylimines multicomponent reactions 

Fast multi-component synthetic reactions guided by interpretable decision trees to suggest new reactions.  

[GitHub: Rxn Mechanisms] 

AI/ML for Analytical Chemistry and Autonomous Instrumentation

1) Neural Networks for Spectroscopy [Paper]

2) Interpretable Machine Learning for Ion-Molecule Reactions [Paper]


LEMON

A new fast data mining method 

LEMON is a framework for rapidly mining structural information from the Protein Data Bank.

[Paper] [GitHub] [Manual]

PADDY

A python package designed for hyperparameter optimization

PADDY is a genetic algorithm for clustering and optimization. It is useful for hyperparameter optimization and it beats Bayesian methods including hyperopt. 

[Website] [GitHub]

DUBS

Software to make benchmarking datasets rapidly

[bioRxiv] -- In Review

SPEAR

Statistical Platform for Elucidating moleculAr Reactivity

SPEAR treats molecules as dynamic graphs to change topology of the molecules during simulation.

[GitHub]

CANDOCK

Computational ANalytics based DOCKing

CANDOCK is our in-house software now integrated with OpenMM to provide Generalized Potential Function to do simulations. It is a fragment based docking method used for small molecule, protein-protein, protein-NA docking. 

[Paper] [GitHub] [Latest version]

CANDO

Computational Analytics of Novel Drug repurposing Opportunities

CANDO is a unique platform to discover therapeutics with higher efficiency, lowered cost, and increased success rates, compared to current approaches.

NEW! CANDO v2 is here! [Paper] [GitHub]

CANDESIGN

Computational ANalytics for DESIGN - A software package capable of novel drug design.

CANDESIGN is a tool for differential target based design that utilizes the docking functionality provided by CANDOCK. It allows for automated drug design to take place after docking has occurred via CANDOCK using the fragments generated with CANDOCK.