Machine Learning in the Identification of Key Residues of Variants in the Interaction of ACE2 and TMPRSS2 Proteins with Spike of SARS-CoV2

• Grant number: 22/02782-5
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: May 01, 2022
• End date: December 31, 2022
• Field of knowledge: Biological Sciences - Biophysics - Molecular Biophysics
• Principal Investigator: Silvana Giuliatti
• Grantee: Ana Carolina Damasceno Sanches
• Host Institution: Faculdade de Medicina de Ribeirão Preto (FMRP). Universidade de São Paulo (USP). Ribeirão Preto , SP, Brazil

Abstract

It is known that the binding affinity between the Spike (S) protein and the angiotensin-converting enzyme 2 (ACE2) and transmembrane serine protease II (TMPRSS2) receptors is one of the main determining factors in the replication rate of SARS-CoV -2 (Severe Acute Respiratory Syndrome coronavirus-2) and that such interactions directly affect the clinical condition of the patient. SARS-CoV-2 is an RNA virus and has a higher mutation rate. This characteristic is very well represented by the variants that have emerged in the last two years of the pandemic. Studies suggest that genetic polymorphisms present in coding regions of ACE2 and TMPRSS2 targets may affect susceptibility, severity and clinical outcome of patients affected by this disease. However, how these mutations and polymorphisms, found in different populations, contribute to improve the stability and affinity of interaction between the SARS-CoV2-ACE2 and SARS-CoV2-TMPRSS2 complexes is not fully understood. Normal modes analysis, as well as molecular dynamics are examples of approaches employed in an attempt to achieve full understanding of the process. These methods generate large amounts of data, but do not allow extracting important features, such as regions or residues in the interaction of structures that can significantly contribute to the interaction between proteins. Some of these differences may be subtle. Thus, interpreting and extracting information from these trajectories is not a simple process. Machine learning methods are used in analysis of large amounts of data, as they reduce the dimensionality of the problem. Therefore, it is proposed in this project to use machine learning approaches to reveal the differences in SARS-CoV-2 strains, to investigate the impact of the genetic variability of SARS-CoV-2 and the ACE2 and TMPRSS2 polymorphisms in the interaction region.