tinyIFD: A High-Throughput Binding Pose Refinement Workflow Through Induced-Fit Ligand Docking
tinyIFD: A High-Throughput Binding Pose Refinement Workflow Through Induced-Fit Ligand Docking
复制标题
DOI：
10.1021/acs.jcim.2c01530
复制
发表时间：
2023-05-19
影响因子：
5.6
通讯作者：
Glaser, Jens
中科院分区：
文献类型：
--
作者：
Hsu, Darren J.;Davidson, Russell B.;Glaser, Jens
基于结构的药物发现的关键步骤是预测候选分子是否以及如何与治疗靶点模型结合。然而，大量的蛋白质侧链移动阻止了目前的筛选方法，如对接，从准确预测配体构象，并需要昂贵的改进生产的候选人。我们提出了一个高通量和灵活的配体位姿精化工作流程，称为“tinyIFD”。该工作流程的主要特点包括使用专门的高通量，小系统MD模拟代码mdgx.cuda和主动学习模型动物园方法。我们展示了这个工作流程在一个大型的不同蛋白质靶点测试集上的应用，在前2和前5个姿势中分别找到晶体状姿势的成功率分别为66%和76%。我们还将该工作流程应用于SARS-CoV-2主要蛋白酶（M-pro）抑制剂，其中我们证明了该工作流程中主动学习方面的益处。
A critical step in structure-based drug discovery ispredictingwhether and how a candidate molecule binds to a model of a therapeutictarget. However, substantial protein side chain movements preventcurrent screening methods, such as docking, from accurately predictingthe ligand conformations and require expensive refinements to produceviable candidates. We present the development of a high-throughputand flexible ligand pose refinement workflow, called "tinyIFD".The main features of the workflow include the use of specialized high-throughput,small-system MD simulation code mdgx.cuda andan actively learning model zoo approach. We show the application ofthis workflow on a large test set of diverse protein targets, achieving66% and 76% success rates for finding a crystal-like pose within thetop-2 and top-5 poses, respectively. We also applied this workflowto the SARS-CoV-2 main protease (M-pro) inhibitors, wherewe demonstrate the benefit of the active learning aspect in this workflow.