Alliance for Pandemic Preparedness

February 5, 2020

Machine intelligence design of 2019-nCoV drugs



  • Gao, et al. apply a generative network complex (GNC) machine intelligence approach to identify candidate protease inhibitors for treating 2019-nCoV, including an assessment of two HIV protease inhibitors. 
  • The GNC identifies 3-dimensional (3-D) drug candidates based on elements that include structure generation and property prediction algorithms. Molecule generation includes assessment of properties like binding affinity for known antigenic sites, solubility, similarity to know protease inhibitors, etc. Information on sequence identity and structural similarity between 2019-nCoV and SARS-CoV proteases was the basis of identifying candidates for an initial protease inhibitor dataset for the training set.
  • 15 anti-2019-nCov molecules were generated through the GNC process. A related assessment of the HIV protease inhibitors lopinavir (Aluvia) and ritonavir (Norvir) found them to have some potential (e.g., fit with target protease), but generally scored lower than other protease inhibitors identified.

Gao, et al. (Feb 4, 2020). Machine intelligence design of 2019-nCoV drugs. Pre-Print downloaded on 5 Feb, 2020 from,