Software Installation

Option 1 (highly recommended): use apptainer/singularity

We provide a pre-built Docker image on DockerHub for immediate usage.

Using KaMRaT within apptainer/singularity is highly recommended for users at any level unless software development work is involved, because:

1. Building the container is simple and only requires the dependency of apptainer/singularity which should be pre-installed on most of HPC clusters.
2. Usage of KaMRaT in container ensures better reproducibility of results.
3. Several companion scripts can be easily run within the container, e.g., to construct k-mer count matrix.

In case when installation of apptainer is needed, please refer to their website for instructions.

To build apptainer/singularity image, please run:

apptainer build KaMRaT.sif docker://xuehl/kamrat:latest
    # alternatively to build singularity image: simply replace "apptainer" to "singularity"

Option 2 (advanced user only): build from source

Dependencies

• CMake,
• MLPack >=4.0.0,
• Armadillo,
• Boost-iostreams,
• To use the provided Snakefile that builds input k-mer matrix from fastq files, Snakemake is also required.

If you are installing MLPack with conda, please add the following line into your .bashrc file in the home/ directory before compiling KaMRaT:

export LD_LIBRARY_PATH=/path_to_conda_env/mlpack/lib:$LD_LIBRARY_PATH

Compile source code from GitHub

git clone --recursive https://github.com/Transipedia/KaMRaT.git
cd KaMRaT
cmake . && make -j

Finally, an executable binary file is available as apps/kamrat.