DataSheet_1_Facilitating systems-level analyses of all-cause and Covid-mediated sepsis through SeptiSearch, a manually-curated compendium of dysregulated gene sets.docx

BackgroundSepsis is a dysfunctional host response to infection. The syndrome leads to millions of deaths annually (19.7% of all deaths in 2017) and is the cause of most deaths from severe Covid infections. High throughput sequencing or ‘omics’ experiments in molecular and clinical sepsis research have been widely utilized to identify new diagnostics and therapies. Transcriptomics, quantifying gene expression, has dominated these studies, due to the efficiency of measuring gene expression in tissues and the technical accuracy of technologies like RNA-Seq.
ObjectiveMost of these studies seek to uncover novel mechanistic insights into sepsis pathogenesis and diagnostic gene signatures by identifying genes differentially expressed between two or more relevant conditions. However, little effort has been made, to date, to aggregate this knowledge from such studies. In this study we sought to build a compendium of previously described gene sets that combines knowledge gained from sepsis-associated studies. This would enable the identification of genes most associated with sepsis pathogenesis, and the description of the molecular pathways commonly associated with sepsis.
MethodsPubMed was searched for studies using transcriptomics to characterize acute infection/sepsis and severe sepsis (i.e., sepsis combined with organ failure). Several studies were identified that used transcriptomics to identify differentially expressed (DE) genes, predictive/prognostic signatures, and underlying molecular responses and pathways. The molecules included in each gene set were collected, in addition to the relevant study metadata (e.g., patient groups used for comparison, sample collection time point, tissue type, etc.).
ResultsAfter performing extensive literature curation of 74 sepsis-related publications involving transcriptomics, 103 unique gene sets (comprising 20,899 unique genes) from thousands of patients were collated together with associated metadata. Frequently described genes included in gene sets as well as the molecular mechanisms they were involved in were identified. These mechanisms included neutrophil degranulation, generation of second messenger molecules, IL-4 and -13 signaling, and IL-10 signaling among many others. The database, which we named SeptiSearch, is made available in a web application created using the Shiny framework in R, (available at https://septisearch.ca).
ConclusionsSeptiSearch provides members of the sepsis community the bioinformatic tools needed to leverage and explore the gene sets contained in the database. This will allow the gene sets to be further scrutinized and analyzed for their enrichment in user-submitted gene expression data and used for validation of in-house gene sets/signatures.

背景
脓毒症（Sepsis）是宿主对感染产生的失调性应答。该综合征每年造成数百万例死亡（2017年占全球总死亡人数的19.7%），同时也是重症新冠感染患者死亡的主要原因。在分子与临床脓毒症研究中，高通量测序或“组学”（‘omics’）实验已被广泛用于发掘新型诊断方法与治疗策略。其中，定量检测基因表达水平的转录组学（Transcriptomics）因能够高效检测组织内基因表达，且具备RNA测序（RNA-Seq）等技术的精准性，成为此类研究的主流方法。

目的
现有此类研究多通过鉴定两种及以上相关条件下的差异表达（DE）基因，旨在揭示脓毒症发病机制与诊断基因特征的全新机制性认知。但截至目前，鲜有研究对这类实验所产生的知识进行整合汇总。本研究旨在构建一套整合脓毒症相关研究成果的既往基因集汇编，从而实现与脓毒症发病机制高度相关的基因筛选，并明确脓毒症常见关联的分子通路。

方法
本研究通过PubMed检索以转录组学方法表征急性感染/脓毒症及重症脓毒症（即合并器官衰竭的脓毒症）的相关文献。最终筛选出多项通过转录组学鉴定差异表达基因、预测/预后特征及潜在分子应答与通路的研究。研究收集了各基因集所包含的分子信息，同时整合了对应研究的相关元数据（如用于对比的患者分组、样本采集时间点、组织类型等）。

结果
本研究对74项涉及转录组学的脓毒症相关文献进行了系统性文献整理，最终整合了来自数千名患者的103个独立基因集（涵盖20899个独特基因）及其配套元数据。研究鉴定出基因集中高频出现的基因及其参与的分子机制，其中包括中性粒细胞脱颗粒、第二信使分子生成、白细胞介素4（IL-4）与白细胞介素13（IL-13）信号通路、白细胞介素10（IL-10）信号通路等多种机制。本研究构建的数据库命名为SeptiSearch，基于R语言的Shiny框架（Shiny framework）开发为Web应用程序，访问地址为https://septisearch.ca。

结论
SeptiSearch可为脓毒症研究领域的科研人员提供用于利用与探索数据库内基因集的生物信息学工具。该工具支持对基因集进行进一步审视与分析，以检测其在用户提交的基因表达数据中的富集情况，同时可用于验证内部自建的基因集/特征。