The Vanderbilt Urologic Infection Repository, a Resource for Personalized Clinical Discovery

摘要--资源项目 尿路感染(UTIs)不仅是最常见的泌尿外科疾病之一，也是 最多样化的。既有传统的风险因素，也有新兴的风险因素，常见的临床情景范围 从无症状的菌尿和无并发症的膀胱炎到肾盂肾炎、菌血症和彻头彻尾的尿毒症。 致病微生物同样包括各种各样的机会病原体，还有其他 这些物种的个体菌株之间的(和显著的)基因/表型多样性。在分子水平上， 决定泌尿系健康和尿路感染之间平衡的宿主-病原体因素仍然没有完全确定。 发病机制的概括性模型未能解释现实世界临床上常见的细微差别 实践，限制了医生提供循证和个性化护理的能力。至 为了应对这些挑战，我们寻求创建范德比尔特泌尿系统感染资料库(VUIR)：一个大规模的 (但未确认)收集患者特定的临床信息和配对的微生物分离株， 从我们医疗中心诊断尿培养的常规工作流程中重新调整用途。既是一名信息学家 和生物资源，VUIR将建立在范德比尔特已经到位的独特基础上。这些 包括我们的合成衍生品，我们电子健康记录的匿名镜像，以及microVU，以及 我们诊断实验室的所有无菌现场微生物分离物都是通过这一倡议系统化的 保留用于学术调查。我们现在提议扩大微型VU的活动，以包括范德比尔特令人敬畏的 尿液培养的数量，将储存的微生物(每年数千)与关键的可搜索参数联系起来 来自来源--患者(如人口统计数据、症状、风险因素)以及更广泛的数据 在合成衍生品中。作为一座双向桥梁，VUIR将创造一个机会来解析人类 根据微生物学结果，提供大量野生型微生物菌株的表型组 对于下游实验，均按人类尿路感染表型进行分层。一个特别令人兴奋的应用 VUIR涉及全基因组关联研究，直接将病原体的遗传特征联网 寄主的临床特征。除了构建VUIR之外，我们还将挖掘存储库以选择 全基因组测序中未被充分代表的微生物目标。通过机器学习方法， 这些菌株的多片段基因组特征将与其宿主的临床参数相关，具有 重视大肠杆菌和[1]症状性尿路感染和[2]无症状性尿路感染的全球表型 菌尿(ASB)。区分这些现象学类别的微生物特征仍然很差 尽管UTI的严格分子定义--至少由迄今认为的简化指标--来定义-- VS-ASB对内科医生具有重要的诊断价值，对研究人员具有病理学价值。在灯光下 对于这种复杂性，我们将利用VUIR来协调主机上的生物信息和医疗信息数据 和病原体，作为该计划在临床/基础泌尿外科和相关领域广泛应用的概念验证。

SUMMARY - RESOURCE PROJECT Urinary tract infections (UTIs) represent not only one of the most prevalent urologic pathologies, but also one of the most diverse. With both traditional and emerging risk factors, commonly encountered clinical scenarios range from asymptomatic bacteriuria and uncomplicated cystitis to pyelonephritis, bacteremia, and outright urosepsis. The causative microbial agents likewise include a tremendous variety of opportunistic pathogens, with additional (and significant) genotypic/phenotypic diversity among individual strains of these species. On a molecular level, the host-pathogen factors that dictate the balance between urologic health and UTI remain incompletely defined. Generalized models of pathogenesis fail to account for commonly encountered nuances of real-world clinical practice, limiting the ability of physicians to provide care that is both evidence-based and personalized. To combat these challenges, we seek to create the Vanderbilt Urologic Infection Repository (VUIR): a massive (but de-identified) collection of patient-specific clinical information and paired microbial isolates, repurposed from our Medical Center's routine workflow of diagnostic urine cultures. As both an informatic and biologic resource, the VUIR will build on unique foundations that are already in place at Vanderbilt. These include our Synthetic Derivative, an anonymized mirror of our electronic health records, along with microVU, an initiative through which all sterile-site microbial isolates from our Diagnostic Laboratories are systematically retained for academic inquiry. We now propose to expand microVU activities to include Vanderbilt's formidable volume of urine cultures, linking the banked organisms (many thousand annually) to key searchable parameters from the source-patients (e.g. demographics, symptomatology, risk factors), as well as the broader body of data within the Synthetic Derivative. As a two-way bridge, the VUIR will create an opportunity to parse human phenomes in light of microbiologic results, while providing a tremendous quantity of wild-type microbial strains for downstream experimentation, all stratified by human UTI-phenotypes. One particularly exciting application of the VUIR involves genome-wide association studies (GWAS) that network genetic features of pathogens directly to clinical features of their hosts. In addition to building the VUIR, we will mine the repository to select underrepresented microbial targets for whole-genome sequencing. Through a machine-learning approach, the multi-partite genomic features of these strains will be correlated to their hosts' clinical parameters, with an emphasis on Escherichia coli and the global phenotypes of [1] symptomatic UTI and [2] asymptomatic bacteriuria (ASB). The microbial features that distinguish these phenomenological categories remain poorly defined—at least by the simplified metrics considered to date—although a rigorous molecular definition of UTI- vs-ASB would carry significant diagnostic value for physicians and pathogenetic value for investigators. In light of this complexity, we will utilize the VUIR to harmonize bioinformatic and medical informatic data across host and pathogen, as proof-of-concept for this program's broad utility in clinical/basic urology and allied fields.