Implementation of the LiBELa Docking Program in Python Language

• Grant number: 23/02134-6
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: July 01, 2023
• End date: June 30, 2024
• Field of knowledge: Biological Sciences - Biophysics - Molecular Biophysics
• Principal Investigator: Alessandro Silva Nascimento
• Grantee: Caio de Jesus de Oliveira
• Host Institution: Instituto de Física de São Carlos (IFSC). Universidade de São Paulo (USP). São Carlos , SP, Brazil

Abstract

In the 14th edition of CASP, one its participants presented a solution with a precision, in many cases, close to experimental precision. It's the neural network AlphaFold2, developed by DeepMind. Since then, the international scientific community started to notice the importance of Artificial Intelligence (AI) in structural biology applications. The amount of available experimental structures and the sheer volume of data structures predicted by AlphaFold2 to sequences in UNIPROT or the PDB archive (currently, there are more than 200.000 available experimentally determined structures in the bank) raises important questions. For instante, how can this amount of structural data contribute to the advancement of our scientific knowledge, or even technological development? How do we integrate this volume of data into useful information for development? These questions can be answered based on multiple perspectives, one of them being molecular design based on the structure of a target macromolecule. In the 80s, the development of docking algorithms allowed a certain automation in the virtual screening of possibly active molecules through the quality evaluation of their interaction with a target macromolecule, or simply, the receptor. The recent advance in computer processing and the recent growth of compound data banks opens the possibility of virtual screening in ultra-large molecular libraries.LiBELa, which stands for Ligand Binding Energy Landscape, is a software developed by the São Carlos Physics Institute that seeks to take the next step in molecular docking by having a mixed approach, one that is both ligand and structure-based. Another important potential improvement to the program would be the use of artificial intelligence in the search for better poses and better estimates in energy interaction. Therefore, in this project, we propose a change in the current LiBELa code, written in C/C++, to the Python language and its future integration with the AI libraries already available in Python. As a result, it would be possible to execute the program in virtual machines, such as the Google Colab environment, without needing to install anything, making its use and dissemination easier. We also aim, with this project, to evaluate the new Python language codes with the usual methods, namely: redocking, cross-docking and enrichment in benchmarks such as DUD-E and DUD-Z.