BugSigDB: A Comprehensive Database of Published Microbial Signatures for Epidemiological Analysis of Microbiome Data

Most microbiome studies report “signatures” of differentially abundant microbial taxa for a disease or exposure of interest, but heterogeneity in how complex methods and results are reported make comparisons between studies difficult. A signature represents statistically different bacterial taxa identified in a given study and represents a summary of a study’s findings. These signatures can be assessed across studies for consistency and agreement. BugSigDB is a manually curated database of microbial signatures collected from published microbiome differential abundance studies. Key information for each study is curated including study design, sample size, participant information, laboratory methods, and statistical methods in a structured, standardized format. BugSigDB is open source and available for researchers to use and contribute to. As of January 11th, 2022 it contains 2,107 microbial signatures curated from 522 published microbiome research articles. We present an applied example of use of BugSigDB gut microbiome signatures for COVID-19. Analysis of 132 signatures from 32 studies of COVID-19 and the gut microbiome reveal significant co-occurrence of several bacterial taxa within the phylum Firmicutes along with Actinobacteria, Bacteriodetes, and Proteobacteria (Figure 1). Across these studies, the genera Bacteroides and Alistipes were statistically found to be in decreased abundance among COVID-19 patients. bugsigdbr, an accompanying open source R/Bioconductor package, provides signatures and accompanying data downloads in a tidy data format. Data can also be accessed via a semantic wiki web interface at https://bugsigdb.org. Together, they allow for efficient systematic analysis of findings from microbiome studies across a variety of diseases and conditions of interest.