README.md

In jburos/biostan: Intro to Stan for BioC2016 workshop

biostan

Materials for BioC-2016 workshop entitled "Introduction to Bayesian Inference using Stan with Applications to Cancer Genomics"

Usage

Most of the material for this package is contained within the vignettes.

To install this package, including vignettes:

devtools::install_github('jburos/biostan', build_vignettes=TRUE, dependencies=TRUE)

To list & review the vignettes, after having installed the package:

vignette(package = "biostan")

Dependencies

This package contains most dependencies in the Requires section of the DESCRIPTION file. The above-mentioned process should install it without any problems under most circumstances.

However, special consideration should be given to the installation of rstan and rstanarm.

clang++ C++ compiler

It is strongly recommended that you install & enable the clang++ C++ compiler on the system in addition to or instead of the more typical g++ compiler. The clang++ compiler will compile Stan models faster and with less memory, so that you can get by with an Amazon AMI that has less RAM.

To install it on Ubuntu, you can do

sudo apt-get install clang++

~/.R/Makevars file

To configure R packages to use clang++, there needs to be a file whose path is ~/.R/Makevars that contains the following:

CXX=clang++
CXXFLAGS=-g -O3

although on Ubuntu the clang++ executable may actually be called something slightly different (ex: clang++-3.4).

In this case the ~/.R/Makevars should use the name of the executable (CXX=clang++-3.4) instead of the above.

install packages from source

Finally, given that v2.10 is expected to be released to CRAN before the workshop (before 6/26), it is recommended for now to install the latest versions of both rstan and rstanarm from github.

The following process should work:

• Install StanHeaders (version 2.10.0) from CRAN because it is hard to install from GitHub (due to its submodules)
• Make sure devtools is installed from CRAN

• Install rstan from GitHub via

  devtools::install_github("stan-dev/rstan", ref = "develop", subdir = "rstan/rstan", build_vignettes = TRUE, dependencies = TRUE) * Install rstanarm from GitHub via

  devtools::install_github("stan-dev/rstanarm", args = "--preclean", local = FALSE, build_vignettes = TRUE, dependencies = TRUE) (Note the additional arg --preclean provided to the install of rstanarm).

jburos/biostan documentation built on May 18, 2019, 9:19 p.m.