A high-throughput FEP pipeline for optimisation of hits into leads for novel drug targets

自由能微扰（FEP）方法越来越多地用于工业中，以指导临床前候选化合物的效价优化。FEP方法非常适合于优化适合于基于结构的药物设计的新型药物靶标的化合物，因为它们可以提供高准确度的效价预测，而不需要活性化合物的先前训练集。然而，FEP的性能因目标类别（例如激酶、GPCR）而异，并且可能受到FEP实施方式的选择的影响。目前，在工业研发环境中有效地部署FEP需要大量的技术专业知识。本研究的目标是开发一个优化的开源FEP实施方案，以实现广泛的药物目标类别的稳健性能。通过开发易于使用的自动化管道，使用BioSimSpace框架进行生物分子模拟（www.biosimspace.org）进行高通量FEP研究，将有助于对领先的FEP实施（可通过Amber，Gromacs，OpenMM等获得）进行系统比较。将使用高性能计算和云计算资源计算蛋白质-配体结合自由能，用于Evotec当前感兴趣的药物靶点的不同数据集。从这些大规模验证研究中吸取的经验教训将推动我们在内部FEP软件SOMD（www.siremol.org）中实施新算法，以提高FEP的准确性，速度和可靠性。通过在Evotec实习期间，学生将把新软件部署到内部计算资源上，并帮助分子建模人员将该方法应用于真实的世界药物发现项目。

Free Energy Perturbation (FEP) methods are increasingly used in industry to guide potency optimisation of preclinical candidate compounds. FEP methods are well suited to optimise compounds for novel drug targets amenable to structure-based drug design as they can deliver high accuracy potency predictions without requiring a prior training set of active compounds. However the performance of FEP varies across target classes (e.g. kinases, GPCRs) and can be affected by the choice of the FEP implementation. Currently significant technical expertise is required to deploy FEP effectively in an industrial R&D setting.The goal of this research is to develop an optimised open source FEP implementation for robust performance across broad range of drug target classes. Systematic comparison of leading FEP implementations (available via e.g. Amber, Gromacs, OpenMM) will be facilitated by development of easy to use automated pipelines for high throughput FEP studies using the BioSimSpace framework for biomolecular simulation (www.biosimspace.org). Protein-ligand binding free energies will be computed using high performance computing and cloud computing resources for a diverse dataset of drug targets of current interest to Evotec. Lessons learned from these large scale validation studies will drive the implementation of novel algorithms in our in-house FEP software SOMD (www.siremol.org) to improve the accuracy, speed and reliability of FEP. Through periods of placement at Evotec the student will deploy the new software onto internal computing resources and help molecular modellers apply the methodology to real world drug discovery projects.