组织蛋白酶L
药物发现
对接(动物)
虚拟筛选
计算生物学
小分子
化学
计算机科学
组织蛋白酶
人工智能
生物
生物化学
酶
医学
护理部
作者
Qi Li,Hao Wang,Weili Yang,Jin‐Kui Yang
标识
DOI:10.1080/17460441.2023.2174522
摘要
Cathepsin L (CTSL) is a promising therapeutic target for metabolic disorders and COVID-19. However, there are still no clinically available CTSL inhibitors. Our objective is to develop an approach for the discovery of potential reversible covalent CTSL inhibitors.The authors combined Chemprop, a deep learning-based strategy, and the Schrödinger CovDock algorithm to identify potential CTSL inhibitors. First, they used Chemprop to train a deep learning model capable of predicting whether a molecule would inhibit the activity of CTSL and performed predictions on ZINC20 in-stock librarie (~9.2 million molecules). Then, they selected the top-200 predicted molecules and performed the Schrödinger covalent docking algorithm to explore the binding patterns to CTSL (PDB: 5MQY). The authors then calculated the binding energies using Prime MM/GBSA and examined the stability between the best two molecules and CTSL using 100ns molecular dynamics simulations.The authors found five molecules that showed better docking results than the well-known cathepsin inhibitor odanacatib. Notably, two of these molecules, ZINC-35287427 and ZINC-1857528743, showed better docking results with CTSL compared to other cathepsins.Our approach enables drug discovery from large-scale databases with little computational consumption, which will save the cost and time required for drug discovery.
科研通智能强力驱动
Strongly Powered by AbleSci AI