Molecular docking, conformational analysis, and 3D QSAR model building for TBK-1 inhibition

• Main Authors: Rallos, Kiana Casenas, Matriano, Ethan Joshua Punzalan
• Format: text
• Language: English
• Published: Animo Repository 2022
• Subjects: Protein kinases—Inhibitors; Chemistry
• Online Access: https://animorepository.dlsu.edu.ph/etdb_chem/12; https://animorepository.dlsu.edu.ph/cgi/viewcontent.cgi?article=1014&context=etdb_chem
• Institution: De La Salle University

TBK-1 inhibition was established to be a favorable target for addressing medical issues such as COVID-19, cancer, obesity, inflammatory diseases, and neurodegenerative diseases, given its vital role in several biological processes involved in cell division, autophagy, innate immune response, inflammation, insulin-dependent pathways, signaling of neurodegenerative diseases, and many others. Approaching this, data-driven computational chemistry, through open source software is advantageous considering its cost-effectiveness, accuracy, and speed in assessing drug candidates as compared to conventional drug discovery techniques, especially since there is still a lack of research studies regarding its application on TBK-1 inhibition. 3D QSAR model development, validation, and implementation, as supplemented by molecular docking, conformational analysis, and alignment, were then utilized guided by parameters that ensure biologically significant ligand binding modes in pursuit of contributing paradigms for facilitating potential drug design and discovery. Three 3D QSAR models were established based on three major aligned Clusters A, B, and C, which represent the chemical scaffolds of substituted 2-amino-5-oxo-5H-chromeno[2,3-b]pyridine-3-carboxylic acid derivatives, 2,4,-diamino-5-cyclopropyl pyrimidine with a phenyl attached at the pyrimidine C2 amine group, and substituted benzimidazoles respectively. 3D RDF descriptors were the most prominent and influential variables in the formulated QSAR models with Cluster A and C having good internal or training set predictability and B with bad or test set predictability, while all Clusters presented bad external predictability based on MAE criteria. However, robustness testing implied all clusters presented good applicability and reliable regression results for all training and test sets. Model application on validation sets also exhibited consistency based on expected applicability and validity of predicted pIC50 activity associated with similar structure and orientation of compounds, which contributed to the reliability and enhanced predictive ability of the constructed models.