Docker to run ipyrad
Installation and user manual is found here:
Tags correspond to the version of ipyrad, currently:
- latest: ipyrad 0.7.9
udocker is a user client tool to run docker containers without docker. It's aimed at running applications in user space without root or admin previledges.
udocker resources can be found in:
- Repository: https://github.com/indigo-dc/udocker
- Installation manual: https://github.com/indigo-dc/udocker/blob/master/doc/installation_manual.md
- User manual: https://github.com/indigo-dc/udocker/blob/master/doc/user_manual.md
- Paper open access: https://doi.org/10.1016/j.cpc.2018.05.021
A summary of using udocker to run ipyrad follows:
mkdir -p ~/bin
curl https://raw.githubusercontent.com/indigo-dc/udocker/master/udocker.py > bin/udocker
chmod u+rx bin/udocker
udocker install
udocker pull lipcomputing/ipyrad:0.7.30
udocker create --name=ipyrad lipcomputing/ipyrad:0.7.30
If you are running with multiple threads or processes, OpenMP or OpenMPI, better performace is acheived with udocker F mode:
udocker setup --execmode=F1 ipyrad
If you have input files to pass to the container to run the application, for
example in ~/input
, you can run:
udocker run -v $PWD/input:/home ipyrad ipyrad -n iptest
Pull the docker image
docker pull lipcomputing/ipyrad
Run ipyrad: get the help usage
docker run lipcomputing/ipyrad ipyrad -h
Or you can run the container in interactive mode:
docker run -it lipcomputing/ipyrad /bin/bash
# ipyrad -h
Other examples in: http://ipyrad.readthedocs.io/outline.html
You can map a directory into the docker container for input and output
If you local dir is pwd
, you can run:
PWD=`pwd`
docker run -it -v ${PWD}:/home/test lipcomputing/ipyrad /bin/bash
# cd /home/test
Such that if you have you input files in your host machine, they will be
available in the directory /home/test
inside the container, as well as
output files produced in that dir by running the application inside the container
they will be available when you exit, in that directory pwd
.
GNU GPL v3
Mario David: [email protected]
LIP Lisbon: http://www.lip.pt
Indigo DataCloud: https://www.indigo-datacloud.eu/