RAPID: Bioinformatic Search for Epitope-based Molecular Mimicry in the SARS-CoV-2 Virus using Chameleon

交叉反应性免疫是一个过程，通过该过程，接种了不相关疫苗的个体或过去从不相关病原体感染中恢复的个体似乎可以免受SARS-CoV-2的感染。个体对病原性感染产生成功免疫应答的关键步骤是人抗体识别并结合来自感染病原体的抗原蛋白的特异性表位（片段）。当抗体偶然结合SARS-CoV-2的表位时，可能发生分子模拟的交叉反应性，因为在结合界面处与其预期的表位具有结构相似性。如果得到证实，针对其他病原体设计的药物和疫苗的快速再利用可以迅速得到验证并应用于当前的大流行。该项目计划使用生物信息学技术来研究分子模仿如何在交叉反应免疫中发挥作用。软件管道将使用Chameleon云计算平台中的高性能计算资源在机器学习框架内运行计算密集型分子动力学模拟，并帮助识别SARS-CoV-2中分子模仿的发生。管道可分为两个主要部分。第一部分涉及从公共数据库如蛋白质数据库（PDB）中获得的已知复合物的结构中提取有用的特征。第二部分涉及从这些特征中构建机器学习模型，以便在SARS-CoV-2中检测分子模拟，如果存在的话。机器学习框架将导致可重复使用的分子模拟模型，并有望帮助疫苗开发。如果成功，该项目有可能：（a）解释住院率和死亡率方面的全球差异;（B）导致迅速改变药物用途，以抗击当前的流行病;（c）推广用于其他病原体;（d）导致更快的疫苗开发;（e）为当前和未来的流行病制定新的生物信息学战略。一个具有计算生物物理学、生物信息学、机器学习、进化生物学、传染病、计算表观遗传学、糖生物学、高性能计算和软件工程将推动这一项目。所有结果将通过项目网站http：//biorg.cs.fiu.edu/lemom提供，包括分子模拟的例子，用于复制实验的软件，以及变色龙的性能基准测试结果。该奖项反映了NSF的法定使命，并被认为值得通过使用基金会的智力价值和更广泛的影响审查标准进行评估来支持。

Cross-reactive immunity is a process by which an individual who was vaccinated by an unrelated vaccine or who has recovered from an infection from an unconnected pathogen in the past is seemingly protected against an infection by SARS-CoV-2. A key step for an individual to mount a successful immune response to a pathogenic infection is for a human antibody to recognize and bind to a specific epitope (fragment) from an antigenic protein from the infecting pathogen. Cross-reactivity by molecular mimicry may occur when an antibody fortuitously binds to an epitope from SARS-CoV-2 because of a structural similarity at the binding interface with the epitope for which it was intended. If verified, a rapid repurposing of drugs and vaccines designed for the other pathogens can be quickly validated and applied to the current pandemic. This project plans to use bioinformatic techniques to investigate how molecular mimicry may play a role in cross-reactive immunity.The software pipeline will use the high-performance computing resources in the Chameleon cloud computing platform to run computationally-intensive molecular dynamics simulations within a machine learning framework and help identify occurrences of molecular mimicry in SARS-CoV-2. The pipeline can be divided into two main parts. The first part involves extracting useful features from structures of known complexes available from public databases such as Protein Data Bank (PDB). The second part involves building machine learning models from these features so that molecular mimicry, if present, can be detected in SARS-CoV-2.The machine learning framework will result in reusable models of molecular mimicry and is expected to assist in vaccine development. If successful, the project can potentially (a) explain global disparities in hospitalizations and death rates; (b) lead to quick repurposing of drugs to fight the current pandemic; (c) be replicated for other pathogens; (d) lead to faster vaccine development; (e) impact development of novel bioinformatic strategies for the current and future pandemics.An interdisciplinary team with expertise in computational biophysics, bioinformatics, machine learning, evolutionary biology, infectious diseases, computational epigenetics, glycobioogy, high-performance computing and software engineering will drive this project. All results will be made available through the project website at: http://biorg.cs.fiu.edu/lemom, including examples of molecular mimicry, software for replicating the experiments, and performance benchmarking results on Chameleon.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.