Swiss Variant Interpretation Platform (SVIPdb)

Introduction

The Swiss Variant Interpretation Platform (SVIPdb) is a framework to collect, curate, review, and share variant annotations. It is the software developed as part of the SPHN infrastructure development project Swiss Variant Interpretation Platform for Oncology (SVIP). We distinguish between the tool and the service by their acronyms SVIPdb and SVIP, respectively.

Main components

This top-level project coordinates three sub-projects that compose SVIP:

1. g2p-aggregator: contains the pipeline (in g2p-aggregator/harvester) that populates the variant database,
2. svip_api: the Django/DRF API server that serves data from the database to the front-end,
3. svip-o-vue: the Vue.js-based web front-end.

docker-compose.yml is the main configuration file for launching the set of containers that compose the complete SVIP system. It includes definitions for a few other essential services, such as the Postgres database and local copy of the Universal Transcript Archive (UTA). These are built from Docker Hub images and thus don't have any local data in this repo.

The top-level git project includes the three sub-projects as git submodules. Typically, we commit to the coordinating project when we want to declare that all the sub-projects are compatible with each other at that point in their respective commit histories. We'll also add tags to the coordinating repo (and probably the sub-repos as well) when we perform official releases.

Prerequisites

You're encouraged to use Docker and docker-compose to run the project; be sure to have these preinstalled.

You must also have Git LFS installed in order to download the default database dumps.

Initial Setup

Clone the project and its submodules

To clone the project and its submodules, you can run the following:

git clone [email protected]:SVIPdb/coordinator.git svip
cd svip
git submodule update --init --recursive

Configure your environment

Make sure you are in the docker group

usermod -aG docker <username>

Make the seqrepo database available locally

sudo mkdir -m 777 -p /usr/local/share/seqrepo
cd seqrepo
make

Create a .env file from the template

cp .env.TEMPLATE .env

Set proper values in your .env file.

...
BIOONTOLOGY_API_KEY=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
ONCOKB_API_KEY=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx

Download the harvester resources

To download the harvester resources you need to have an access token to onkokb as well as for bioontology.

./dc_harvester.sh build harvester
./dc_harvester.sh run --rm harvester /bin/sh -c 'cd /data; make'

Populate the VEP cache

./dc_harvester.sh run --rm harvester /bin/sh -c 'scripts/populate_vep_cache.sh'

Do initial tasks

To initially start the database server, API, and frontend:

1. Copy the contents of .env.TEMPLATE into .env and change the fields in it as necessary.
2. Run docker-compose up -d, which will launch the database server, API server, and static webserver which hosts the frontend.
3. (Optional) If you'd like to load a data dump, execute the following:
   1. docker-compose exec db /bin/bash
   2. cd /backups and ls -lt to review the list of available dumpfiles.
   3. ./restoredump.sh <your_dumpfile> svip_api to load the dumpfile.

To start the harvester:

1. (Optional) Start a screen/tmux session, since this will likely take a while.
2. Run docker-compose run --rm harvester /bin/bash
3. Within the container, edit entrypoint.sh all and then execute it to start the harvesting process.

Notes and other components:

There are currently three environments in which the project can run:

• local, in which the API server runs on port 8085, the front-end runs on 8080, and the database server is accessible for debugging on port 30432.
• dev, which typically is deployed to the svip-dev server.
• external, which typically is deployed to svip-test.

All the environments require that docker and docker-compose are installed.

The deployment folder contains (work-in-progress) scripts that facilitate deployment and synchronization across the three environments.

The db_config folder contains initialization scripts for the postgres database; on first execution, it loads a sample database containing some faked SVIP data and entries from public databases (CIViC, OncoKB, etc.).

Other random folders:

• sql_scratch contains scratch SQL scripts used for ad-hoc analyses.
• notebooks contains some Jupyter notebooks for ad-hoc analyses.
• assets contains artwork and other static assets used in the front-end and promotional.
• svip_db_backups stores snapshots of the database, retrieved from svip-dev.
• kibana contains configuration info for g2p-aggregator's original front-end. (It's essentially an artifact of the g2p-aggregator project and isn't used by SVIP.)
• outputs contains a few dumps of information retrieved from some of the sources for reference purposes.