Rxivist logo

Massive-scale biological activity-based modeling identifies novel antiviral leads against SARS-CoV-2

By Ruili Huang, Miao Xu, Hu Zhu, Catherine Z Chen, Emily M. Lee, Shihua He, Khalida Shamim, Danielle Bougie, Wenwei Huang, Matthew Hall, Donald Lo, Anton Simeonov, Christopher P. Austin, Xiangguo Qiu, Hengli Tang, Wei Zheng

Posted 27 Jul 2020
bioRxiv DOI: 10.1101/2020.07.27.223578

The recent global pandemic caused by the new coronavirus SARS-CoV-2 presents an urgent need for new therapeutic candidates. While the importance of traditional in silico approaches such as QSAR in such efforts in unquestionable, these models fundamentally rely on structural similarity to infer biological activity and are thus prone to becoming trapped in the very nearby chemical spaces of already known ligands. For novel and unprecedented threats such as COVID-19 much faster and efficient paradigms must be devised to accelerate the identification of new chemical classes for rapid drug development. Here we report the development of a new biological activity-based modeling (BABM) approach that builds on the hypothesis that compounds with similar activity patterns tend to share similar targets or mechanisms of action. In BABM, compound activity profiles established on massive scale across multiple assays are used as signatures to predict compound activity in a new assay or against a new target. We first trained and validated this approach by identifying new antiviral lead candidates for Zika and Ebola based on data from ~0.5 million compounds screened against ~2,000 assays. BABM models were then applied to predict ~300 compounds not previously reported to have activity for SARS-CoV-2, which were then tested in a live virus assay with high (>30%) hit rates. The most potent compounds showed antiviral activities in the nanomolar range. These potent confirmed compounds have the potential to be further developed in novel chemical space into new anti-SARS-CoV-2 therapies. These results demonstrate unprecedented ability using BABM to predict novel structures as chemical leads significantly beyond traditional methods, and its application in rapid drug discovery response in a global public health crisis. ### Competing Interest Statement The authors have declared no competing interest.

Download data

  • Downloaded 678 times
  • Download rankings, all-time:
    • Site-wide: 45,414
    • In cell biology: 1,867
  • Year to date:
    • Site-wide: 50,144
  • Since beginning of last month:
    • Site-wide: 75,061

Altmetric data

Downloads over time

Distribution of downloads per paper, site-wide