This repository contains source files for the "Orchestrating Spatial Transcriptomics Analysis with Bioconductor" book.
The development version of the book is available at: https://lmweber.org/OSTA/
The book is built by the GitHub Actions workflows in .github/workflows/
.
The GitHub Actions workflows are split into two files:
depbuild.yml
installs dependencies and stores these in a Docker image (triggered only if there are changes toDESCRIPTION
ordeps.Dockerfile
)biocbook.yml
builds the book and runs checks (triggered by changes to any other files)
A new public version of the book is deployed if changes are pushed to devel
branch.
Please send pull requests into sandbox
branch (instead of devel
). This will trigger GitHub Actions to build the book and run checks, but will not deploy a new version of the book. The maintainers will then merge into devel
to deploy a new version of the book after reviewing and accepting the pull request.
To build a local version of the book, run the following in an R session in the inst/
directory (after installing all dependencies):
quarto::quarto_render(cache = TRUE)
To compile a complete fresh local build, delete all existing build files by deleting the following directories and files: inst/.quarto/
, inst/docs/
, inst/index_cache/
, any files and directories except .qmd
files and images/
directory in inst/pages/
(e.g. directory names ending with _cache/
or _files/
, .md
files, and .rds
files). You may also need to empty the BiocFileCache
cache directory if any data files stored in the OSF data repository have changed.
In each analysis / method chapter, we aim to include the following:
- brief background and overview
- mention any relevant benchmark papers
- demonstrate 1-2 key methods available in Bioconductor using code examples
- (optional) mention any other important methods in text (but keep concise)
Workflow chapters are longer and are intended to demonstrate a comprehensive workflow for a given data type / technological platform.
Please keep any added file sizes (e.g. images) as small as possible, and/or commit only code files.