R package for analyzing specificity in ecological data
by John L. Darcy Updated 27 JUL 2023
The word "specificity" is often used in the context of host-parasite interactions. For example, a parasite may be "specific" to a narrow phylogenetic group of hosts. This is phylogenetic specificity. However, in this R package I define specificity more broadly: a species has specificity to a variable when that species occupies a limited range of that variable. A species may have specificity to elevation if it can only be found within a narrow elevational band. With this software, you can compute specificity for many species at once, to different types of variables including vectors (e.g. elevation), matrices (e.g. pairwise geographic distance), ontologies, and phylogenies. This R package can be used to analyze patterns of ecological specificity, particularly in microbiome data. Please check out the tutorial vignette in the vignette folder of this repository, which will walk you through the use of this software.
install.packages("remotes")
remotes::install_github("darcyj/specificity")
pre-compiled binaries are available at https://github.com/darcyj/specificity_builds. Download the build that matches your OS, and install it from terminal using R CMD INSTALL specificity_blahblahblah.tgz.
specificity needs a few R packages to run. They should be installed automatically when you run the above. In case they aren't, you'll need:
Rcpp(In order to compile the C++ code inspecificity. For users on Apple OS X, you may be asked by Rstudio to install "command line developer tools". If you're on OS X and not using Rstudio, you'll need to runxcode-select --installfrom terminal beforehand. If installing from Rstudio and asked whether to compile from source, you'll get best results if you say "yes")ape(for phylogenetic stuff)fields(for geographic stuff)
For parallel processing you'll need to be using Linux or OS X, since R's parallelism doesn't work well on Windows. If you're using Windows, just set n_cores to 1 for any given paralellized function. OR if you're on Windows 10, running R through Windows Subsystem for Linux may give better multithreading results - please let me know if you try this.
A full tutorial for specificity can be found in the vignette folder of this repository. To find it, just go to github.com/darcyj/specificity and click on the folder called "vignette". In there you'll find both a .pdf (for everybody) and .rmd (for rstudio users) that will walk you through the package.
Documentation for individual functions can be obtained within R using ?fun where fun is the function you're interested in. There is also a full .pdf documentation file in this repository named specificity.pdf, which is a standard glossary of all functions in the package.
If you'd like to contribute to this project, let me know! Right now, it would be nice to have:
- more visualizations
- vignette for advanced use (using your own null model)
- vignette for integration of common data formats (i.e. phyloseq)
- more builds at https://github.com/darcyj/specificity_builds
Send me an email if you'd like.