Quick Start
⏱ tutorial time: 10 minutes
Welcome to the 10 Minutes to Isabl guide! This tutorial will walk you through installation, meta data registration, data import, and automated data processing.
Isabl is composed of a patient centric relational model, a web-based metadata architecture, and a command line client.
Make sure your installation doesn't require
sudo to run docker and docker-compose. Otherwise you will have issues running this demo. Check docker run hello-world runs without problem. If you have permissions issues, see how to run docker as-non-root user.Let's start by clone the demo:
git clone https://github.com/papaemmelab/isabl_demo.git --recurse-submodules
cd isabl_demo
Make sure the
git submodule folders isabl_api and isabl_cli are not empty. If they are, probably the --recurse-submodules flag didn't work. As a workaround, run:
git submodule update --recursive --initNext, source a simple initiation profile:
source .demo-profile
If you are redoing the tutorial, we recommend to remove the demo directory and clone it again:
chmod -R u+w isabl_demo && rm -rf isabl_demo
Also remove the Docker volume:
docker volume rm isabl_demo_local_postgres_data
This installation relies on docker private images. Please make sure with the isabl team admins that you have proper access to them.
Build and run the application (this might take a few minutes):
Before running
docker-compose build, specify your platform architecture if it's different from the standard intel linux/amd64, i.e. you have an Apple M1/M2: export DOCKER_DEFAULT_PLATFORM=linux/arm64
If you are not sure, you can check your platform by running:
uname -m
demo-compose build
Now we can run the application in the background:
demo-compose up -d
You can type
demo-compose down to stop the application. And use demo-compose logs -f in a new session to see logs.Create a superuser by running
demo-django createsuperuser, for example using credentials: username=admin password=admin [email protected]demo-compose, demo-django, and demo-cli are simple wrappers around docker-compose - check them out. The isabl_demo directory was bootstrapped using cookiecutter-isabl, a proud fork of cookiecutter-django! Many topics from their guide will be relevant to your project.Creating a project in Isabl is as simple as adding a title. You can also specify optional fields:
Hover over the menu and click in the + icon to add a new project.
We need to let
isabl-cli know where can the API be reached, and what CLI settings we should use (if you are using demo-compose, these variables are already set):# point to your API URL
export ISABL_API_URL=http://localhost:8000/api/v1/
# set environment variable for demo client
export ISABL_CLIENT_ID="demo-cli-client"
Now we can create a client for
isabl-cli:# create and update the client object
demo-cli python3.6 ./assets/metadata/create_cli_client.py
# check the file used to create the client
cat ./assets/metadata/create_cli_client.py
Before we create samples, let's use
isabl-cli to add choices for Center, Disease, Sequencing Technique, and Data Generating Platform:demo-cli python3.6 ./assets/metadata/create_choices.py
We will use Excel submissions to register samples through the web interface. To do so, the demo project comes with a pre-filled metadata form available at:
open assets/metadata/demo_submission.xlsm
When prompted to allow macros, say yes. This will enable you to toggle between optional and required columns. By the way, Isabl has multiple mechanisms for metadata ingestion! Learn more here.
Now let's proceed to submit this excel form. First open the directory:
open assets/metadata
And drag the
demo_submission.xlsm file into the submit samples form:
Click the + button in the project panel header to add new samples.
We can also review metadata at the command line:
isabl get-metadata experiments --fx
Expand and navigate with arrow keys, press e to expand all and E to minimize. Learn more at
fx documentation. Use --help to learn about other ways to visualize metadata (e.g. tsv).For this particular demo, we wanted to create a sample tree that showcased the flexibility of Isabl's data model. Our demo individual has two samples, one normal and one tumor. The tumor sample is further divided into two biological replicates (or aliquots), with two experiments conducted on the second aliquot:
A data generation process tree that resulted in 4 sequencing experiments (or ultimately bams), produced from two samples of the same individual.
Although not required for this tutorial, you are welcome to checkout the RESTful API documentation at: http://localhost:8000/api/v1/ or https://isabl.github.io/redoc/.
Given that
isabl-cli will move our test data, let's copy original assets into a staging directory:mkdir -p assets/staging && cp -r assets/data/* assets/staging
Now let's import the genome:
isabl import-reference-genome \
--assembly GRCh37 \
--genome-path assets/staging/reference/reference.fasta
We can also import BED files for our demo Sequencing Technique:
isabl import-bedfiles \
--technique DEMO_TECHNIQUE \
--targets-path assets/staging/bed/targets.bed \
--baits-path assets/staging/bed/baits.bed \
--assembly GRCh37 \
--species HUMAN \
--description "Demo BED files"
Check that import was successful:
isabl get-bed DEMO_TECHNIQUE # retrieve BED file
isabl get-reference GRCh37 # retrieve reference genome
By means of the
--data-id flag, the command get-reference also allows you to retrieve the indexes generated during import. To get a list of available files per assembly run:isabl get-reference GRCh37 --resources
Next step is to import data for the samples we just created:
isabl import-data \
-di ./assets/staging `# provide data location ` \
-id identifier `# match files using experiment id` \
-fi identifier.contains "demo" `# filter samples to be imported `
Add
--commit to complete the operation.Retrieve imported data for the normal to see how directories are created:
isabl get-data -fi sample.identifier "demo normal"
The front end will also reflect that data has been imported.
Isabl is a language agnostic platform and can deploy any pipeline. To get started, we will use some applications from isabl-io/apps. Precisely we will run alignment, quality control, and variant calling. Applications were previously registered in client object. Once registered, they are available in the client:
isabl apps-grch37
First we'll run alignment (pass
--commit to deploy):isabl apps-grch37 `# apps are grouped by assembly ` \
bwa-mem-0.7.17.r1188 `# run bwa-mem version 0.7.17.r1188 ` \
-fi tags.contains data `# filter using tags, feel free to try others `
Note that if you try to re-run the same command, Isabl will notify you that results are already available. If for some reason the analyses fail, you can force a re-run using
--force.Now we can retrieve bams from the command line:
isabl get-bams -fi sample.individual.identifier "demo individual"
We can also visualize aligned bams online:
Insert
2:123,028-123,995 in the locus bar, that's were our test data has reads. Learn more about default BAMs in the Writing Applications guide.
Although the BAM file is an output of the bwa-mem analysis, Isabl enables registering default bams to an experiment. Thus a link is available in the sample panel.
isabl apps-grch37 qc-data-0.1.0 -fi identifier.icontains demo --commit
This quality control application has defined logic to merge results at a project and individual level. Upon completion of analyses execution, Isabl automatically runs the auto-merge logic:
A short message is displayed at the end of the run indicating merge analyses are being run.
Isabl-web can render multiple types of results, in this case we will check at HTML reports. Results for our
qc-data application are available at an experiment, individual, and project level. In this example we are looking at the project-level auto-merge analysis:
A project level Quality Control report. Can you find the Experiment and Individual-level reports?
Up until now we've run applications that are linked to one experiment only. However, analyses can be related to any number of target and reference experiments. For example this implementation of Strelka uses tumor-normal pairs. Before you can run this command you will need to retrieve the system id of your experiments, let's try:
isabl get-metadata experiments -f system_id
Now insert those identifiers in the following command:
isabl apps-grch37 strelka-2.9.1 \
--pairs {TUMOR 1 ID} {NORMAL ID} `# replace tumor 1 system id and normal system id` \
--pairs {TUMOR 2 ID} {NORMAL ID} `# replace tumor 2 system id and normal system id` \
--pairs {TUMOR 3 ID} {NORMAL ID} `# replace tumor 3 system id and normal system id`
You can retrieve registered results for the analysis, for instance the indels VCF:
isabl get-results -fi name STRELKA --result-key indels
To find out what other results are available use:
# app-primary-key can be retrieved from the frontend
isabl get-results --app-results {app-primary-key}
# when writing this tutorial, the app key for strelka was 5
isabl get-results --app-results 5
Furthermore, you can get paths for any instance in the database using
get-paths:isabl get-outdirs -fi name STRELKA
Lastly, lets check the indels VCFs through the web portal:
To finalize the tutorial, we'll use Isabl as an SDK with
ipython:# we'll access ipython using the cli container
demo-cli ipython
Then lets check the output directories of Strelka:
import isabl_cli as ii
# retrieve analyses from API using filters
analyses = ii.get_analyses(name="STRELKA")
# list the strelka ouput directories
for i in analyses:
!ls {i.storage_url}/strelka
analysis = analyses[0]
# get the target experiment or tumor
target = analysis.targets[0]
# print the parent sample class
print(target.sample.category)
# see available fields
print(target.keys())
Learn about writing applications:
Last modified 16d ago