A Phylodynamic Artificial Intelligence framework to predict evolution of SARS-CoV-2 variants of concern in Immunocompromised persons with HIV (PhAI-CoV)

总结 美国是全球受持续严重急性呼吸系统疾病影响最严重的国家， 冠状病毒2型（SARS-CoV-2）大流行。有效疫苗的供应最初放缓， 减少新的感染，减少2019年严重冠状病毒病（COVID-19）病例的发生率，住院治疗 负担和死亡。不幸的是，疫苗的犹豫，以及新的，高传染性变种的出现， 值得关注的挥发性有机化合物（VOCs），如德尔塔变种，已迅速成为美国的主导之一， 未接种疫苗和疫苗突破病例，在7月至8月期间造成了新的疫情激增 2021年，可能是一个持续存在的问题，阻碍了根除流行病的努力。虽然数据增加， 最近，艾滋病病毒感染者（PWH）的死亡率和更差的临床结局在某种程度上是模棱两可的， 调查显示，威尔斯亲王医院的病人较没有免疫力的病人， 功能障碍此外，虽然大多数人在2 - 4周内有效地清除了SARS-CoV-2，但一些感染报告 在免疫抑制的个体中，已经显示出宿主内出现多突变变体， 与免疫逃避有关，特别是在持续感染的情况下。拟议项目的总体目标 是研究SARS-CoV-2基因组在HIV感染背景下的宿主内进化， 评估SARS-CoV-2 VOC出现和可能性的动态和人工智能框架 （PhAI-CoV）在免疫功能低下的PWH中。假设威尔斯亲王医院的SARS-CoV-2感染可导致 增强的病毒变体的进化，可以有效地跟踪通过免疫动力学分析和预测， 通过人工智能算法检测挥发性有机化合物。为了验证这一假设，我们制定了三个具体目标， 调查三个互补但独立的问题。我们将使用一个特征明确的PWH队列， 并严格收集了迈阿密SARS-CoV-2合并感染患者的纵向数据和样本， 佛罗里达是美国艾滋病毒和SARS-CoV-2感染负担最高的城市之一。具体目标1， 将招募并保留n = 120例急性SARS-CoV-2感染的PWH，以及n = 120例匹配的对照组， 急性SARS-CoV2感染，但没有艾滋病毒，并研究COVID-19疾病的严重程度如何因艾滋病毒状态而异， 这取决于SARS-CoV-2疫苗接种史和感染变体。在具体目标2中，我们将调查内部- 宿主SARS-CoV-2在整个感染过程中的演变，以评估SARS-CoV-2感染的可能性 导致持续的宿主内进化，从而导致新的病毒变体的出现。在特定 目标3，我们将开发一种人工智能算法，可以预测新变种的可能性， 挥发性有机化合物了解艾滋病毒感染中出现的SARS-CoV-2变异体的进化情况， 评估菌株感染性和/或致病性增加的可能性将为以下方面提供重要工具： 在一般人群中发生传播之前规划和实施公共卫生措施。

Summary The United States (US) is the most affected country worldwide by the ongoing Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) pandemic. The availability of effective vaccines had initially slowed down new infections, reducing incidence of severe coronavirus disease 2019 (COVID-19) cases, hospitalization burden, and deaths. Unfortunately, vaccine hesitancy, and the emergence of new, highly transmissible variants of concern (VOCs), such as the Delta variant that has rapidly become the dominant one in the US among both non-vaccinated and vaccine breakthrough cases, have caused a new dramatic epidemic surge in July-August 2021, and are likely to be an ongoing problem hindering epidemic eradication efforts. Although data on increased mortality and worse clinical outcome in people with HIV (PWH) with COVID-19 is somewhat equivocal, recent surveys indicate that PWH has a higher likelihood of severe disease or death than patients without immune dysfunction. Moreover, while most people effectively clear SARS-CoV-2 in 2-4 weeks, several reports of infection in immunosuppressed individuals have shown intra-host emergence of multi-mutational variants, some at sites linked to immune evasion, especially in case of persistent infection. The overarching goal of the proposed project is to investigate SARS-CoV-2 genomes intra-host evolution in the context of HIV infection by developing a phylodynamic and artificial Intelligence framework to assess the emergence and likelihood of SARS-CoV-2 VOC (PhAI-CoV) in immunocompromised PWH. The hypothesis is that SARS-CoV-2 infection in PWH can result in enhanced evolution of viral variants that can efficiently be tracked by phylodynamic analysis and predicted to be VOCs by artificial intelligence algorithms. To test such a hypothesis, we developed three specific aims that will investigate three complementary, albeit independent, issues. We will use a well-characterized cohort of PWH and rigorously collected longitudinal data and samples from patients with SARS-CoV-2 co-infection in Miami, Florida, one of the cities with the highest HIV and SARS-CoV-2 infection burden in the US. In Specific Aim 1 we will recruit and retain n=120 PWH with acute SARS-CoV-2 infection, as well as n=120 matching controls with acute SARS-CoV2 infection but without HIV, and study how COVID-19 disease severity differs by HIV status, depending on SARS-CoV-2 vaccination history and infecting variant. In Specific Aim 2, we will Investigate intra- host SARS-CoV-2 evolution throughout the duration of infection to assess the likelihood of SARS-CoV-2 infection in PWH to result in sustained intra-host evolution leading to the emergence of novel viral variants. In specific Aim 3, we will develop an artificial intelligence algorithm that can predict the likelihood of new variants to be VOCs. Understanding the evolutionary scenarios of SARS-CoV-2 variants emergence within HIV infection and evaluating the probability for increased strain infectivity and/or pathogenicity will provide a crucial tool for planning and implementing public health measures before transmission occurs in the general population.