Skip to content

Latest commit

 

History

History
406 lines (283 loc) · 16.7 KB

README.md

File metadata and controls

406 lines (283 loc) · 16.7 KB
Raw

Intro To Docker for Public Health Bioinformatics Training

Table of Contents

Prerequisites

  • GitHub Account
    • If you don't already have a GitHub account, please go to github.com and register (it's free!)
    • Once you have registered ensure you are signed into your account on github.com
  • Navigate to GitPod & start a Workspace:
    • https://gitpod.io/
    • Select "Continue with GitHub", authorize Gitpod to access your GitHub account, and sign in with your credentials
    • Start New Workspace
    • Set the "Context URL": https://github.com/theiagen/docker-builds
    • Select the default Editor and default Class, click "Continue"
  • 🚨 Once you're finished working, make sure to shut down your workspace to save your allotted free hours in GitPod 🚨. There are a few ways to stop your workspace from running:
    • In GitPod environment (VSCode interface), click on the orange Gidpod in bottom left corner. Select "Stop Workspace"
    • OR you can navigate to https://gitpod.io/, find your workspace in the list, click on the three-dot button, and select "Stop"

Docker basics

Check to ensure docker is installed, look at help options for docker and docker pull. Run the following commands in the Terminal:

docker --version
docker --help
command -v docker
docker pull --help

The hello world exercise for docker! What happens when you run this command?

docker run hello-world

docker images = command used to list all docker images available on your computer. What docker images are available on your computer now?

docker images

Downloading docker images

Let's download another docker image to practice with.

We're going to download a docker image from this docker hub repository: https://hub.docker.com/r/staphb/ncbi-datasets. Run the following command to download the StaPH-B docker image that contains the NCBI datasets command-line tool. More info on the command line tool datasets can be found here

docker pull staphb/ncbi-datasets:14.20.0

Now run the command docker images - what has changed?

QUIZ: What command would I use to download the StaPH-B docker image for the mlst software? (hint: see https://hub.docker.com/r/staphb/mlst/tags)

Answer can be found here. Click to show answer: 
docker pull staphb/mlst:latest

Docker images vs containers

When you ran the command docker run hello-world earlier, docker started a container using the docker image called hello-world, ran some things, and then the container automatically exited. The container did not delete itself, and we can see that by running the command docker ps --all. Run the command:

docker ps --all

You can tell docker to automatically delete the container after exiting with the docker run --rm flag. Let's try it out. Run the following command which will download and run a command in a new docker container:

docker run --rm ubuntu:focal echo "hello from inside the container!"

Now, run docker ps --all - do we see a container listed for the command we just ran? We should not see it listed, as long as the docker run --rm flag was included.

Now try the same command without the --rm flag:

docker run ubuntu:focal echo "hello from inside the container!"

# look at the list of containers again:
docker ps --all

Running docker images

Interactive mode: You can launch into a container interactively, and be given a pseudo-shell from which you can run commands. You will need to use the docker run -it flags to do this:

# your command prompt will change after launching interactive mode
$ docker run -it staphb/ncbi-datasets:14.20.0
Unable to find image 'staphb/ncbi-datasets:14.20.0' locally
14.20.0: Pulling from staphb/ncbi-datasets
2ab09b027e7f: Already exists 
d6391dffcf79: Pull complete 
4f4fb700ef54: Pull complete 
eaa7a46260b5: Pull complete 
05a7178bde59: Pull complete 
Digest: sha256:4441c4556698fce13a2612889d467435eb8cec096565233da0e7b74fbae4a3fb
Status: Downloaded newer image for staphb/ncbi-datasets:14.20.0
root@d1c3f33f5fc9:/data

# now run the following commands:
pwd
ls 
cd /
ls

# you can exit the interactive container with the 'exit' command
exit

Non-interactive mode: The default method of running a container does not launch an interactive terminal, but rather, a command or set of commands is passed into the container. The format is as follows:

# format
docker run <name-of-docker-image>:<tag> <command>

# example
docker run ubuntu:focal echo "hello!"

File permissions & passing files in and out of containers

File permissions

By default, when you launch a container with docker run, any commands run will be run as the root linux user. Thus, any files created will be owned by the root user, limiting the downstream use of these files by non-root linux users. Non-root users may encounter file permission errors if attempting to delete or edit files owned by root.

To avoid these issues, you can pass in your linux user ID (UID) and group ID (GID) so that the container runs with your user and group IDs. The flag to pass in user and group IDs is docker run --user <user-id>:<group-id> or docker run -u <user-id>:<group-id> for short

To test let's run the StaPH-B NCBI datasets docker container and create some files. Run the following commands:

# change directories so that we have a relatively clean working environment
cd training/GT-Comp-Genomics-Docker-for-PH-Bioinformatics-Jan2024/

# launch the container in interactive mode
docker run -it staphb/ncbi-datasets:14.20.0

# create a file called "test.txt"
touch test.txt

# look at the permissions for this file
ls -lh

# see what the current user you're logged in as
whoami

# exit the container with the command 'exit'
exit

Passing files in and out of containers

Docker containers have filesystems that are completely separate and isolated from your host computer's filesystem. You can mount a "volume" between your host computer's filesystem and the container filesystem to pass files in and out.

Files created inside of containers are not automatically exported from the container so we must use the --volume <path-on-host>:<path-in-container> or -v <path-on-host>:<path-in-container> (short hand). You must tell docker which path to mount from your host system, and the path on the container's filesystem to mount the volume to. It's OK if the path inside the container does not exist, it will be created.

If my present working directory ($PWD) is /home/curtis_kapsak, then I can mount my $PWD to /data in the container's filesystem like so:

docker run --volume $PWD:/data ubuntu:focal <command>

Let's pass a file into a container. First make an test text file, and then pass into the container:

# create the text file which contains the text "this is a test
echo "this is a test">text.txt

# mount your PWD to /data inside the container and list the contents of /data in the container filesystem
docker run --volume $PWD:/data ubuntu:focal ls /data

# same as before, but cat the test.txt file from inside the container
docker run --volume $PWD:/data ubuntu:focal cat /data/text.txt

Now let's create a file inside the docker container, and then save it to our host computer for usage later:

# create a file inside the container using the "touch" command, writing to /data in the container
docker run --volume $PWD:/data ubuntu:focal touch /data/file-made-inside-container.txt

# let's check and make sure that new file exists on our host computer filesystem:
ls

# what happens when we don't mount a volume and create a file inside the container?
docker run ubuntu:focal touch file-made-inside-container-NO-VOLUME.txt

# is this new file present on our host filesystem? Run this command to check:
ls file-made-inside-container-NO-VOLUME.txt

Week 1 Exercise 1

Now that we've covered some of the basics of docker on the command line, let's download a Klebsiella pneumoniae genome FASTA file from NCBI.

End goal: Use NCBI datasets to download a genome FASTA file - Klebsiella pneumoniae

We downloaded the docker image called earlier, so now let's use it to download the FASTA file.

We're going to download this genome: https://www.ncbi.nlm.nih.gov/data-hub/genome/GCF_000240185.1/

Let's launch the datasets docker image interactively, and use the flags you learned about earlier:

### launch the container in interactive mode ###
# --rm is to remove the container after it exits (i.e. delete the container)
# -v is for mounting my filesystem to the container filesystem
# -u is for passing in my user ID and group ID
# -it is 2 flags for launching interactive mode
docker run --rm -v $PWD:/data -it staphb/ncbi-datasets:14.20.0

# run the datasets tool to download the FASTA file (along with some other metadata)
datasets download genome accession GCF_000240185.1

# unzip the .zip archive
unzip ncbi_dataset.zip

# exit the container
exit

# take a peek at the top of the FASTA file
head -n 4 ncbi_dataset/data/GCF_000240185.1/GCF_000240185.1_ASM24018v2_genomic.fna

Week 1 Exercise 2

Goal: Run kleborate on FASTA file for subtyping, serotyping, virulence and AMR prediction

Now that we have downloaded our FASTA file, let's launch into the container in interactive mode, and run kleborate

# download the docker image "staphb/kleborate:2.3.2-2023-05" and launch interactive mode
docker run --rm -v $PWD:/data -it staphb/kleborate:2.3.2-2023-05

# run kleborate using the FASTA file as input (be patient, this step may take a few minutes to run....)
kleborate --all -o results.tsv -a ncbi_dataset/data/GCF_000240185.1/GCF_000240185.1_ASM24018v2_genomic.fna

# exit the container
exit

# view results of Kleborate
cat results.tsv

Week 2 Exercises

Let's build some docker images! In today's exercise we are going to practice building docker image in various ways. Let's tweak the dockerfiles and see how those changes impact the final docker image.

NCBI datasets

Let's build the most recent version of NCBI datasets using the dockerfile located here: ncbi-datasets/14.20.0/Dockerfile

Reminder - the basic docker build command structure:

docker build --tag <name>:<tag> <directory-with-dockerfile>

Now with real values filled in, run the following command:

docker build --tag ncbi-datasets:14.20.0 /workspace/docker-builds/ncbi-datasets/14.20.0/

What happened when we ran this command?

  • Docker daemon read & interpreted the dockerfile. It also ensured correct syntax & format.
  • The base image ubuntu:jammy was downloaded if not already present on computer.
  • Daemon began running subsequent Dockerfile instructions (RUN, WORKDIR, etc) in order through to the end of the dockerfile.
  • Docker image was given a sha256 hash & we received a message saying docker image was named & built successfully.

When we ran the previous command, it built all stages of the dockerfile. BUT we can tell docker to build to a specific stage using docker build --target <stage>. Let's try building to the test stage specifically:

# note the new option '--target test'
docker build --target test --tag ncbi-datasets:14.20.0-test-stage /workspace/docker-builds/ncbi-datasets/14.20.0/

That command finished running almost instantaneously, why?

Answer can be found here. Click to show answer:

Answer: The image has already been built previously (with last docker build command we ran) and thus the layers are "cached" or stored locally to be re-used. No need to spend time & resources building an image when it has already been built!


Now lets build to only the `app` stage, as this will be the final docker image that we share via dockerhub, quay, whatever container registry. This means we will use `--target app` which will skip building the layers in the `test` stage: 
# note the new option '--target app'
docker build --target app --tag ncbi-datasets:14.20.0-app-stage /workspace/docker-builds/ncbi-datasets/14.20.0/

SPAdes

The SPAdes dockerfile is a bit more complex as it's test stage runs the SPAdes software on a toy dataset and assembles a plasmid sequence.

Let's build the most recent version of SPAdes, but this time start with only building the app stage. We are going to use the dockerfile located at spades/3.15.5/Dockerfile:

# note the new option '--target app'
docker build --target app --tag spades:3.15.5-app-stage /workspace/docker-builds/spades/3.15.5/

Now let's build all the way through the test stage and see the test assembly process run:

# note the new option '--target test'
docker build --target test --tag spades:3.15.5-test-stage /workspace/docker-builds/spades/3.15.5/

Experiment

Let's try changing a few things in the dockerfile to see the effect of the changes.

Hypothetical (& false) scenario 1 - SPAdes produced a .zip output file and my bioinformatics pipeline needs to unzip this file to extract the contents. I have to install the unzip software in the SPAdes docker image to accomplish this

Steps:

  1. Open up the SPAdes dockerfile in the GitPod/VSCode editor
  2. Navigate to the top of dockerfile and look for the apt-get install step on lines 18-22.
  3. Add a return and unzip \ after line 21. The entire RUN layer should look like this:
RUN apt-get update && apt-get install --no-install-recommends -y python3 \
 python3-distutils \
 wget \
 pigz \
 unzip \
 ca-certificates && \
 apt-get autoclean && rm -rf /var/lib/apt/lists/* && \
 update-alternatives --install /usr/bin/python python /usr/bin/python3 10
  1. Once added, save the file (CTRL + S) and let's rebuild the image with a new tag (spades:3.15.5-test-unzip-added):
# same command as before, but now with an updated dockerfile
docker build --target test --tag spades:3.15.5-test-stage-unzip-added /workspace/docker-builds/spades/3.15.5/
  1. Launch an interactive container to see if unzip is actually installed:
# launch into interactive mode in container
docker run -ti spades:3.15.5-test-stage-unzip-added

# pull up unzip help options
unzip --help

Hypothetical scenario 2 - I don't think the SPAdes test (spades.py --test) is good enough, I want to make sure it works on some real data. Let's try using SPAdes to assemble some real bacterial WGS data

The current test is sufficient, but let's add another layer to the dockerfile to bolster the test stage. We are going to add some lines to download some real bacterial WGS data and assemble in the test stage. We are going to download some E. coli WGS data, Illumina paired-end, and assemble the genome with SPAdes. Here's the dataset on ENA: https://www.ebi.ac.uk/ena/browser/view/SRR6903006

Steps:

  1. Open up the SPAdes dockerfile in the GitPod/VSCode editor
  2. Navigate to bottom of dockerfile and add these lines of code:
# download test FASTQ files from ENA
# run SPAdes on FASTQ files
RUN wget ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR690/006/SRR6903006/SRR6903006_1.fastq.gz && \
wget ftp://ftp.sra.ebi.ac.uk/vol1/fastq/SRR690/006/SRR6903006/SRR6903006_2.fastq.gz && \
spades.py -t 4 --isolate --only-assembler -1 SRR6903006_1.fastq.gz -2 SRR6903006_2.fastq.gz -o spades-output
  1. Once added, save the file (CTRL + S) and let's rebuild the image:
# same command as before, but now with an updated dockerfile
docker build --target test --tag spades:3.15.5-test-stage-added-test /workspace/docker-builds/spades/3.15.5/

Adding this test will ensure the robustness of the docker image, but the tradeoff is that it takes longer to run the test. The assembly process can take a while (5-15 min or longer), especially if the input dataset is large.

Resources