Skip to content
Switch branches/tags
Go to file

Accelerating Antimicrobial Discovery with Controllable Deep Generative Models and Molecular Dynamics

This work will be published in Nature Biomedical Engineering on March 11, 2021


De novo therapeutic design is challenged by a vast chemical repertoire and multiple constraints, e.g., high broad-spectrum potency and low toxicity. This project proposes CLaSS (Controlled Latent attribute Space Sampling) - an efficient computational method for attribute-controlled generation of molecules, which leverages guidance from classifiers trained on an informative latent space of molecules modeled using a deep generative autoencoder. We screen the generated molecules for additional key attributes by using deep learning classifiers in conjunction with novel features derived from atomistic simulations.


  • The amp_gen.yml lists are the required dependencies for the project.
  • Use amp_gen.yml to create your own conda environment to run this project. Command: conda-env create -f amp_gen.yml


Phase 1: Autoencoder (VAE/WAE) Training

  • ./ This will run with default config from Since cfg.runname=default the output goes to output/default and tb/default.
  • python --tiny 1 for fast testing with default config file.
  • Additionally, one could explicitly run the individual scripts as follows:
    • python --phase 1

    • python --config_json output/dir/config_overrides.json

Phase 2: CLaSS (Controlled Latent attribute Space Sampling)

  • python --config_json output/default/config_overrides.json --samples_outfn_prefix samples --Q_select_amppos 0


Related Visualization Tools


Please cite the following articles:

  title={Accelerating Antimicrobial Discovery with Controllable Deep Generative Models and Molecular Dynamics},
  author={Das, Payel and Sercu, Tom and Wadhawan, Kahini and Padhi, Inkit and Gehrmann, Sebastian and Cipcigan, Flaviu and Chenthamarakshan, Vijil and Strobelt, Hendrik and Santos, Cicero dos and Chen, Pin-Yu and others},
  journal={arXiv preprint arXiv:2005.11248},
  title={CogMol: Target-specific and selective drug design for COVID-19 using deep generative models},
  author={Chenthamarakshan, Vijil and Das, Payel and Hoffman, Samuel C and Strobelt, Hendrik and Padhi, Inkit and Lim, KW and others},
  journal={arXiv: 2004.01215},


source code for "Accelerating Antimicrobial Discovery with Controllable Deep Generative Models and Molecular Dynamics"




No releases published


No packages published