2026-02-042026-02-042025-10-06ALMEIDA, Aguinaldo Pantoja de. Hybrid computational approach with gnina and molecular dynamics in the search for inhibitors against marburg virus. Orientador: Claudomiro de Souza de Sales Júnior. 2025. 102 f. Dissertação (Mestrado em Engenharia Elétrica) - Instituto de Tecnologia, Universidade Federal do Pará, Belém, ano de defesa. Disponível em:https://repositorio.ufpa.br/handle/2011/17945 . Acesso em:.https://repositorio.ufpa.br/handle/2011/17945Molecular modeling has become a crucial tool in studying interactions between drugs and vira targets, allowing the prediction of binding mechanisms and potential therapeutic applications. In this context, four drugs were examined — Ivermectin, Nafamostat, Lopinavir, and Remdesivir — for their potential interactions with the Marburg virus (MARV), exhibiting effective interactive characteristics that suggest a strong possibility of being repurposed for therapeutic approaches against MARV. The effectiveness of these interactions was evaluated through molecular docking, which determines the positions of ligands around the target protein and measures them based on binding energy. The results provided binding energies obtained using GNINA, which employs Convolutional Neural Networks, ranging from –8.74 kcal/mol to –10.54 kcal/mol for Ivermectin,–7.37 kcal/mol to –10.46 kcal/mol for Lopinavir, –7.16 kcal/mol to –10.00 kcal/mol for Nafamos-tat, and –6.56 kcal/mol to –9.59 kcal/mol for Remdesivir. The molecular complexes identified in the docking simulations, together with their respective binding affinities, demonstrated the establishment of interactions between amino acids in the catalytic regions of macromolecular structures and the pharmacological compounds. Furthermore, the electrostatic potential distribution was analyzed to identify promising regions for viral attack. Molecular dynamics simulation approaches were applied, focusing on the structural components of the virus, specifically the transmembrane glycoprotein and the VP24, VP35, VP40, and nucleoprotein (NP) proteins. Using the GROMACS 2025 modules, the results showed that the ligands — the tested drugs Ivermectin, Nafamostat, Lopinavir, and Remdesivir — maintained interactive characteristics over time. From the molecular dynamics simulations, root mean square deviation (RMSD) data of atomic positions were obtained, yielding ranges from 0.25 to 2.0 Å for Ivermectin, 0.25 to 2.37 Å for Lopinavir, and 0.25 to 7.0 Å for Nafamostat. RMSD values below 2–3 Å suggest good structural integrity and stable interactions throughout the simulation. In the results obtained from molecular dynamics, the intermolecular hydrogen bonds showed significant differences compared to the RMSD data, displaying variations in the number of hydrogen bonds during the simulation. These variations were influenced by the ligand’s approach, which helps eliminate non-natural conformations during structural adjustment. Hydrogen bond analysis demonstrated that the amino acids located in the active region of VP40 established significant interactions with the three drugs. Ivermectin exhibited stronger interactions with VP24 and VP30, while Lopinavir interacted with GP2, VP24, and VP30. Nafamostat showed a high number of hydrogen bonds with all the proteins. The repurposing of these drugs against MARV may lead to potential viral antagonists, which, if confirmed by experimental studies, could contribute to reducing the mortality caused by MARV.enAttribution-NonCommercial-NoDerivatives 4.0 Internationalhttp://creativecommons.org/licenses/by-nc-nd/4.0/Docagem molecular MARVMapa de potencial Eletrostático,Dinâmica molecularIvermectina,NafamostatLopinavirMolecular DockingElectrostatic Potential MapMolecular Dynamics MARVIvermectinNafamostatMARV - Marburg virusDissertaçãoCNPQ::ENGENHARIAS::ENGENHARIA ELETRICAINTELIGÊNCIA COMPUTACIONALCOMPUTAÇÃO APLICADA