cpgen-parallel


The package <code>cpgen</code> makes use of shared memory multi-threading using
OpenMP. R is of single-threaded nature, hence almost the entire package is written
in C++. The package offers a variety of functions that lets you control and check
the number of threads that are being used by the functions of the package.
Internally every function uses the global variable <code>cpgen.threads</code> which is stored in
<code>options()$cpgen.threads</code>.
The value can be changed using the function <code>set_num_threads()</code>. When the package is loaded
in an R-session <code>cpgen.threads</code> will be set to the value returned by <code>get_max_threads()</code> which
is a wrapper for the OpenMP-header function <code>omp_get_max_threads()</code>

Parallelization

Frequently used methods in genomic applications with emphasis on parallel computing (OpenMP).
At its core, the package has a Gibbs Sampler that allows running univariate linear
mixed models that have both, sparse and dense design matrices. The parallel sampling method
in case of dense design matrices (e.g. Genotypes) allows running Ridge Regression or BayesA for
a very large number of individuals. The Gibbs Sampler is capable of running Single Step Genomic Prediction models.
In addition, the package offers parallelized functions for common tasks like genome-wide
association studies and cross validation in a memory efficient way.

Parallelization: Multithreading using `cpgen`

Description

Arguments

Details

See Also