Introduction
OmicABELnoMM is a tool for running Genome-Wide Association Studies (GWASn) on large multi-omics datasets in a computationally efficient manner. The tool is specifically designed to run linear regression on thousands of phenotypes (e.g. metabolomics data) for thousands of individuals and millions of (imputed) genotypes within a reasonable time frame (think days instead of months or years).
Requirements
Hardware
OmicABELnoMM requires a modern high-end computer with a multi-core processor. Currently, OmicABELnoMM has experimental support for MPI, which means that the tasks can be distributed over different compute nodes in a cluster.
Software
OmicABELnoMM is designed to run on Linux (it won’t run on MS Windows; well, we actually never tested that). In terms of libraries, OmicABELnoMM depends on LAPACKe and a BLAS library. There are several choices for a BLAS library: OpenBLAS, AMD’s ACML or the Intel MKL. For the calculation of p-values OmicABEL depends on the Boost library.
If you would like to use/test the MPI functionality, you will need either the OpenMPI or the MPICH2 library.
Installation
Because OmicABELnoMM was designed and developed with High-Performance Computing (HPC) in mind, it works most efficiently when compiled from the source code, including the libraries it depends on, on the machine it will be run. This will allow the compiler to choose the most optimal settings that match your hardware (mainly the processor).
Unfortunately, the process of compiling from source, while not prohibitively complicated, is not trivial. As an alternative to compiling both OmicABELnoMM and its dependencies from source you can opt for installing existing packages provided in the repositories of your Linux distribution. However, this will come at a cost: performance will be lower than if you compile the package and its libraries for your own architecture.
Support
Community support for OmicABELnoMM can be found at the GenABEL forum. Feel free to ask your questions there and to contribute with your own knowledge and skills.
Commercial support for OmicABELnoMM can be obtained via PolyOmica. Typical things PolyOmica can help you with are installing OmicABELnoMM on your server or cluster, or the development of custom features.
Contributing
OmicABELnoMM is a free software project, we therefore welcome contributions of all kind. Development takes place on GitHub, where you will find the source code and the list of bugs. Feel free to submit your bug reports or feature requests there, or, even better, fork the project and start fixing those bugs. Development discussions take place on the GenABEL development mailing list. We are looking forward to working with you!
Licence
OmicABELnoMM is free software licensed under the GNU General Public
License v3. A copy of the licence can be found in the file
COPYING.