Install the groHMM latest version using:

devtools::install_github("coregenomics/[email protected]")

If the above command fails with something like Error in loadNamespace(name) : there is no package called ‘devtools’ , install the devtools package first:

install.packages("devtools")

One can see where more unit test coverage is needed from the web reports of codecov. Otherwise, one can check Travis CI for notes and warnings to fix.

To tackle this and more, it's best not to rely on the web CI services (which are very thorough but take an hour to run) and to instead run the tests, coverage and checks locally (which take tens of seconds) as explained below.

Fork or clone the git repository, enter the git directory, then install the dependencies:

source("https://bioconductor.org/biocLite.R")
install.packages("devtools")
devtools::install(dependencies = TRUE)

Run the unit tests with:

devtools::test()

The web integration reports take about 20 minutes to generate; the pacing item being Travis CI.

One can therefore run covr and the checks locally:

library(covr)
cov <- package_coverage()
cov
zero_coverage(cov)

Besides testing and coverage, one can catch a broader range of quality issues using R's standard check followed by BiocCheck:

devtools::check(build_args="--no-build-vignettes")
BiocCheck::BiocCheck(".")

Strictly speaking, Bioconductor's development process requires using a recent R daily build. There are a few different approaches for compiling from source: for example one can use the package manager to install build-time dependencies; on Debian one could run apt-get build-dep r-base. However I suggest using spack and environmental modules instead to allows one to easily switch between R versions and keep a separate R library for each installation.

cd
git clone https://github.com/llnl/spack.git
# Add spack to your PATH.
echo >> ~/.bashrc
echo "export PATH=`readlink -e spack/bin`:\$PATH" >> ~/.bashrc
source ~/.bashrc
# Find the version of R-devel you want at
# https://cran.r-project.org/src/base-prerelease/ and add it to the
# package file.  In my case I needed to add the line:
#
#     version(
#         'date-2018-04-07',
#         url=('https://cran.r-project.org/src/base-prerelease/'
#              'R-devel_2018-04-07_r74551.tar.gz'))
spack edit r
spack install --no-checksum r@2018-04-07  # Change to your version

Install environmental-modules to load our new r module:

aptitude install environment-modules

Note the version of r that you have loaded. We will add the version to our R-devel launcher script.

spack find r

Create a file /usr/local/bin/R-devel:

#!/bin/bash
. /etc/profile.d/modules.sh
. ~/spack/bin/spack/setup-env.sh
spack load r@2017-04-07  # Change version to match `spack find r`
exec R "$@"

In emacs ess-mode as long as R-devel is in your PATH, one can launch it with M-x R-devel as your associated R shell. If you use RStudio or RStudio server, to use R-devel as your interpreter with environmental variables or Rprofile hacks you're on your own <3