MOLECULAR DOCKING AND MOLECULAR DYNAMICS FOR DETERMINING GENE INHIBITORS THAT HAVE POTENTIAL AS RISK FACTORS FOR DOWN SYNDROME

• Main Author: Oleta Palit, Oscar
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/86944
• Institution: Institut Teknologi Bandung

This study aims to identify the interactions between potential inhibitor compounds and genes associated with Down syndrome risk. A total of 15 inhibitor compounds and 15 overexpressed genes in individuals with Down syndrome were identified. Among the compounds, 11 are derivatives of imidazo[1,2-a]pyridine, while 4 are flavonoid catechin derivatives. Binding affinities between the compounds and genes were analyzed using molecular docking methods with YASARA software and the Autodock VINA algorithm. The results revealed three complexes with the best binding affinities: DYRK1A-8o (-11.2 kcal/mol), BACE2-8o (-11.1 kcal/mol), and TFAM-8o (-11.8 kcal/mol). The stability of these complexes was further analyzed using molecular dynamics simulations with YASARA and the AMBER14 force field for 50 nanoseconds. Molecular dynamics analysis showed that the DYRK1A complex is the least stable, as indicated by higher hydrogen bond counts, Solvent Accessible Surface Area (SASA), Radius of Gyration (Rg), Root Mean Square Deviation (RMSD), and Root Mean Square Fluctuation (RMSF) compared to the other complexes. The BACE2 and TFAM complexes demonstrated better stability, with RMSD and RMSF values below 3 Å. Machine learning analysis was subsequently performed to predict RMSD values and evaluate their accuracy against actual RMSD values from molecular dynamics simulations. Four methods were utilized: Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN), and Artificial Neural Network (ANN). Prediction visualization indicated that the DYRK1A complex closely resembled the actual condition, despite having the highest Root Mean Square Error (RMSE) compared to the BACE2 and TFAM complexes. Among the methods, CNN achieved the lowest average RMSE. The most promising inhibitor compound was identified as compound 8n, due to its low binding affinity and stability in the BACE2 and TFAM complexes.