bugphyzz

bugphyzz is a harmonized database of bacterial and archaeal physiological traits (aerophilicity, Gram stain, shape, growth temperature, G+C content, habitat, and about 30 more), collected from curated and programmatic sources, extended to uncharacterized taxa by ancestral state reconstruction, and delivered as an R/Bioconductor package for building microbial signatures for bug set enrichment analysis. Site: https://waldronlab.io/bugphyzz/

Installation

if (!requireNamespace("BiocManager", quietly = TRUE))
    install.packages("BiocManager")
BiocManager::install("bugphyzz")            # release
BiocManager::install("waldronlab/bugphyzz") # devel from GitHub

Usage

library(bugphyzz)
bp <- importBugphyzz()                       # list of data.frames, one per attribute
sigs <- makeSignatures(bp[["aerophilicity"]], taxIdType = "NCBI_ID", taxLevel = "genus")
getTaxonSignatures("562", bp)                # the 21 signatures containing E. coli

importBugphyzz() downloads the released annotations from Zenodo (the release pinned in its version default; pass another DOI, or version = "devel" for the weekly export) and keeps imputed annotations only where they passed cross-validation (mccThreshold, r2Threshold). See vignette("bugphyzz") for the data schema and an enrichment example, and vignette("sources") for the sources.

How the data are produced

The package is the user-facing end of a pipeline spread over several repositories.

Step	Where	What
1. Curation	Google Sheets listed in inst/extdata/spreadsheet_links.tsv	Hand-curated annotations, one sheet per physiology, with the controlled vocabulary in inst/extdata/attributes.tsv and the sources in inst/extdata/attribute_sources.tsv.
2. Programmatic sources	bugphyzzExports, sources/	Scripted imports kept under version control with the script and pinned upstream version: Bergey's Manual genus abstracts (LLM extraction), Madin et al. 2020, BacDive (API), NCBI RefSeq assemblies (genome size, G+C, coding genes). Corrections go in each source's overrides.tsv, never in the output.
3. Raw import	this package, physiologies()	Reads steps 1 and 2 into one tidy table per physiology; for curators and the export, not for users.
4. Harmonization, conflict resolution, propagation	taxPPro (R package) and bugphyzzExports inst/scripts/export_bugphyzz.R	Resolves conflicting annotations by source confidence, propagates strain annotations to species, runs ancestral state reconstruction on the LTP tree (phytools for discrete, castor for numeric), applies the validation thresholds, writes CSV and GMT files. A GitHub Action runs it weekly and commits to devel.
5. Cross-validation	taxPProValidation	10-fold cross-validation of the reconstruction per attribute value; its validation_summary.tsv is copied to inst/extdata/ here and decides which imputed annotations ship. Re-run when sources change.
6. Release	Zenodo	Versioned deposits of the exported files; importBugphyzz() pins one release as its default and is updated with each deposit.
7. Analyses and paper	bugphyzzAnalyses, bugphyzz-manuscript (private)	Articles reproducing the manuscript figures; manuscript issues.

Other related repositories: taxoniq (dated NCBI taxonomy releases for taxid lookups), BugSigDB and bugsigdbr (published signatures used in the enrichment analyses), MicrobiomeBenchmarkData (benchmark datasets).

Contributing

• Data corrections: open an issue here with the curation label, naming the taxon, attribute, source and the sheet or sources/ file; or edit the sheet directly if you have access. Corrections to programmatic sources go in that source's overrides.tsv in bugphyzzExports.
• New sources: add a directory under sources/ in bugphyzzExports following its README (script, pinned version, README, manifest row), then a row in attribute_sources.tsv here.
• New attributes: add the values to attributes.tsv, the physiology to spreadsheet_links.tsv (a sheet link, or none for a source-only attribute), signature bins to thresholds.tsv for numeric attributes, and the attribute to the list in export_bugphyzz.R.
• Code: pull requests get an automatic review; R CMD check and BiocCheck run in CI. Keep NEWS.md current with user-facing changes.