Installation of the kmer_tools conda evironment - KamilSJaron/k-mer-approaches-for-biodiversity-genomics GitHub Wiki

These instructions assume you have already conda installed (e.g. miniconda, anaconda, mamba). To check your conda is function you can execute

conda --help

if you don't see the conda help page, do not proceed further.

Setting up conda

Set up a new conda envirnment and activate it

conda create -n kmer_tools
conda activate kmer_tools

note that you can save condas to alternative places by specifying something like --prefix=/full/path/to/conda/env/kmer_tools instead of the parameter -n.

Now we can proceed by installing individual k-mer tools. Once you will execute following lines, it will be fetching the data from the repositories for a moment and prompt you to confirm you want to download the 400MB of packages (contains both python and R, therefore the relatively big size)

conda install -c bioconda numpy smudgeplot kmc kat

And finally, install Genomescope 2.0 (that is unfortunately not on conda). First, download it, then enter the directory and start an R session

git clone https://github.com/tbenavi1/genomescope2.0
cd genomescope2.0
R

now, within R run following commnads to install 2 dependencies and the genomescope package

install.packages('minpack.lm', repos = "http://cran.us.r-project.org")
install.packages('argparse', repos = "http://cran.us.r-project.org")
install.packages('.', repos=NULL, type="source")
q() # quit R

and the one last step is to copy the genomescope execution script to your conda (again in your shell)

cp genomescope.R "$CONDA_PREFIX"/bin

Now you are done, so you can delete the downloaded repository

cd .. && rm -rf genomescope2.0

verifying the installation

Check all the tools run smoothly by printing their help pages

smudgeplot.py -h
genomescope.R -h
kat --help # note -h would drop a bit cryptic error "Throw in function Mode parseMode"
kmc -h

debugging

If any of the tools do not work as expected (e.g. showing an error message instead of the help page), try to first google the error. If that does not work, contact you local computational support - it will be most likely a cluster/system specific error and they will be able to help you faster than anyone else. Then you can try to post on our slack if anyone had the same issue, and if not, you will be left with no other choice than contacting the developers of individual tools.