Artificial inteligence driven drug repositioning strategy for COVID-19

• Grant number: 20/05369-6
• Support type: Regular Research Grants
• Duration: May 01, 2020 - April 30, 2022
• Field of knowledge: Biological Sciences - Microbiology
• Principal Investigator: Fabio Trindade Maranhão Costa
• Grantee: Fabio Trindade Maranhão Costa
• Home Institution: Instituto de Biologia (IB). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Assoc. researchers: Carolina Horta Andrade ; José Luiz Proença Módena ; Rafael Elias Marques Pereira Silva
• Associated research grant: 17/18611-7 - Development of new tools for search and validation of molecular targets for therapy against Plasmodium vivax, AP.TEM

Abstract

The arrival of COVID-19 in Brazil in March 2020, with its rapid dissemination, exponential growth of cases and the lethality of infected patients, has generated panic in the population, unstructured public and private health systems and negatively impacted the economy of Brazil and around the world. Although the search for an effective vaccine is urgent, its development should require at least one year and a half until its approval. Thus, the need to discover a treatment that can be approved quickly for use in infected patients is evident. Drug repositioning offers a potentially faster approach to identifying drugs already approved for use in humans. Artificial intelligence (AI) is a cutting-edge area of knowledge that allows for the rapid identification of potentially active compounds with appropriate pharmacokinetic and toxicological properties, leading to faster, higher success rate and lower cost in the drug discovery process. In this context, in order to transform drug discovery from a slow, sequential and high-risk process to a fast, integrated model with diminished risk of failure, this project aims to develop an integrated platform based on artificial intelligence to accelerate the drug repositioning for the treatment of COVID-19 with potential for rapid clinical development, through the integration of high-performance computing with chemical and biological data and use of emerging biotechnological and experimental technologies. To this end, we have added different expertise from researchers at the Institute of Biology at UNICAMP and other research institutions in Brazil and abroad and will use a multidisciplinary approach that will involve the development and application of these artificial intelligence tools to guide and accelerate the repositioning of drugs in clinical use. This will be based on inhibition of virus entry into the host cell through inhibition of interactions between human Spike viral and ACE-2 proteins and antiviral activity tests on cell culture in a biological containment laboratory level 3. As preliminary data, in our recently approved paper for publication in the journal Drug Discovery Today, we performed a docking-based virtual screening of all FDA-approved drugs (~2,400 drugs) into SARS-CoV-2 spike glycoprotein complexed with the host ACE2 protein using Schrödinger Glide software, with a grid centered at the spike-ACE2 interface. After docking, the drugs were reordered using machine learning models developed using the Bayesian algorithm and ECFP6 fingerprints descriptors for SARS-CoV phenotypic data. At the end of this approach, 25 approved drugs were selected to be submitted to in vitro testing against SARS-CoV-2 in Vero cells. (AU)