A cloud-based WGS platform for routine surveillance of plasmid-borne carbapenem resistance

质粒传播的碳青霉烯类耐药性已成为医院环境中的严重全球威胁。作为 医院面临着治疗选择不断减少的情况，追踪患者的动向非常紧迫 碳青霉烯类抗性质粒和相关的抗性基因。虽然全基因组测序已 质粒具有巨大的诊断潜力，但由于缺乏清晰、明确的结构，因此带来了一系列独特的挑战 并含有广泛的重复遗传元件。此外，质粒通常会经历复杂的 在环境压力（例如持续暴露于 抗生素。最后，通过水平基因转移的质粒移动经常导致进一步的 多样化，主要来自特定宿主的选择压力。由于这些挑战，跟踪 通过全基因组测序进行质粒的移动通常需要艰苦的手动操作 生物信息学方法。 在此提案中，我们首先关注携带 KPC 基因的肠杆菌科质粒， 开发一个自动化诊断平台，使医院能够密切监测耐药性 质粒。我们的平台将为感染控制单位提供一个可搜索的基于云的数据库，其中包含所有药物 耐药质粒、它们与各种菌株的关联以及医院内的各个进入点。 我们的方法使用两种不同的比对方法进行质粒鉴定：一种方法 旨在快速识别质粒的通用内容；另一种方法旨在独特地 量化抗性基因周围的转座子区域。我们的提案目标是：1）开发质粒 使用全局和转座子生物标记序列量化质粒的框架； 2）开发一个 质粒-菌株比对框架，将搜索新菌株的质粒生物标记序列，然后 鉴定含有相似全局生物标志物序列的其他菌株； 3) 开发转座子比对 框架将搜索携带转座子序列的 KPC 基因的新菌株，然后识别其他菌株 具有相似生物标志物序列的菌株。 我们的方法在一家医院针对两个密切相关的质粒进行了测试。通过首先建立 关键生物标志物，我们能够从 NCBI 快速识别出 240 多个遗传了相同特征的菌株 四年来，跨越三个物种的质粒。我们的算法还能够识别关键的 HGT 其中一个质粒的事件，该事件在一项已发表的研究中被独立记录。 我们基于云的质粒诊断框架将通过处理超过 4,000 来自 NCBI 的肠杆菌科菌株。将使用手动生物信息学方案进行验证。

Plasmid-borne carbapenem resistance has become a serious global threat in hospital settings. As hospitals face a dwindling number of treatment options, there is a great urgency to track the movement of carbapenem-resistant plasmids and the associated resistant genes. While whole genome sequencing has great diagnostic potential, plasmids pose a unique set of challenges as they lack a clear, well-defined structure and contain a wide range of repetitive genetic elements. Additionally, plasmids routinely undergo complex rearrangements during patient colonization, driven by environmental pressures such as continual exposure to antibiotics. Finally, plasmid movements through horizontal gene transfer frequently results in further diversification, predominantly from host-specific selection pressures. As a result of these challenges, tracking the movement of plasmids through whole genome sequencing often requires painstaking manual bioinformatics approaches. In this proposal, with an initial focus on Enterobacteriaceae plasmids that carry the KPC gene, we look to develop an automated diagnostics platform that will allow hospitals to closely monitor drug-resistant plasmids. Our platform will provide infection control units a searchable cloud-based database of all their drug- resistant plasmids, their association with various strains, and the various points of entry within the hospital. Our approach uses two different alignment methodologies for plasmid identification: one methodology is designed to rapidly identify the generic content of the plasmid; the other methodology is designed to uniquely quantify the transposon regions around resistance genes. Our proposal aims are: 1) Develop a plasmid framework that quantifies plasmids with both global and transposon biomarker sequences; 2) Develop a plasmid-strain alignment framework that will search new strains for plasmid biomarker sequences and then identify other strains that contain similar global biomarker sequences; 3) Develop a transposon-alignment framework that will search new strains for KPC-gene carrying transposon sequences and then identify other strains with similar biomarker sequences. Our approach was tested against two closely-related plasmids at a single hospital. By first establishing key biomarkers, we were able to rapidly identify over 240 strains from NCBI that had inherited the same plasmids over a four-year period and across three species. Our algorithm was also able to identify a key HGT event for one of the plasmids, which was independently documented in a published study. Our cloud-based plasmid diagnostics framework will be implemented by processing over 4,000 Enterobacteriaceae strains from NCBI. Validations will be made using manual bioinformatics protocols.