Dockerizing your pipeline

Modified on Fri, 7 Apr, 2023 at 12:57 PM

Docker containers provide an isolated environment in order to both distribute and run software tools with ease. We recommend placing each major step of the pipeline into its own docker container, to be run in a step-wise fashion. This page provides instructions for containerizing your pipeline with docker (information on saving docker images remotely can be found here).

Step 1: create the Dockerfile

In a folder containing any necessary scripts or files, create and open a file called "Dockerfile":

nano Dockerfile

Specify a base image to build upon (e.g. an empty ubuntu image), by adding to the file:

FROM ubuntu

Install any required dependencies(e.g. install, make, gcc, and git):

RUN apt-get update -qq \
 && apt-get install -y make gcc \
 && apt-get install -y git

(use the -y flag to ensure installation continues through any prompts)

If installing a tool from github:

RUN git clone https://github.com/alexdobin/STAR.git

For R scripts requiring external packages, install these packages with:

RUN R -e "install.packages('BiocManager')"
RUN R -e "BiocManager::install('DESeq2')"

Specify the program to run when container is started

ENTRYPOINT ["/STAR-2.7.10b/bin/Linux_x86_64/STAR"]

Exit and save the Dockerfile

Step 2: build the docker image

specify the Dockerfile and a tag (-t) by naming the image and assigning a version number (e.g. ":0.0.1")

docker build -f Dockerfile -t image_name:0.0.1 .

The docker image tag can be changed in order to push to your dockerhub repo with:

docker tag image_name:0.0.1 dockerhub_user_name/image_name:0.0.1

Step 3: push to dockerhub

docker push dockerhub_user_name/image_name:0.0.1

(check here for more information on saving docker images)