HiGlass Server

The HiGlass Server supports HiGlass and HiPiler by providing APIs for accessing and uploading tiles generated by Clodius.

Note: that the HiGlass Server itself only provides an API, and does not serve any HTML.

Installation

Prerequirements:

• Python >=v3.6

Docker

The easiest way to run HiGlass with HiGlass Server is with Docker. More information is available at higlass-docker or check out the Dockerfile.

This project also includes a Dockerfile in the docker-context directory that can be used to run a locally checked out copy of higlass-server as follows:

docker build -t higlass-server -f docker-context/Dockerfile .
docker run -d --cap-add SYS_ADMIN --device /dev/fuse --security-opt apparmor:unconfined --name higlass-server higlass-server

Manually

To install HiGlass Server manually follow the steps below. Note we strongly recommend to create a virtual environment using Virtualenvwrapper for example. Skip step 2 if you don't work with virtual environments.

git clone https://github.com/higlass/higlass-server && cd higlass-server

Manually with virtualenvwrapper

mkvirtualenv -a $(pwd) -p $(which python3) higlass-server && workon higlass-server
pip install --upgrade -r ./requirements.txt
python manage.py migrate
python manage.py runserver

Manually with conda

conda env create -f environment.yml
conda activate higlass-server
python manage.py migrate
python manage.py runserver

To enable the register_url api endpoint, HiGlass depends on a project called httpfs to cache external url files. Tests depend on this process running. Set it up as follows:

pip install simple-httpfs

mkdir -p media/http
mkdir -p media/https
simple-httpfs media/http
simple-httpfs media/https

Or simply use ./unit_tests.sh.

Uploading Files

Although there is an API endpoint for uploading files, but it is more direct to use a manage.py script:

COOLER=dixon2012-h1hesc-hindiii-allreps-filtered.1000kb.multires.cool
HITILE=wgEncodeCaltechRnaSeqHuvecR1x75dTh1014IlnaPlusSignalRep2.hitile

wget -P data/ https://s3.amazonaws.com/pkerp/public/$COOLER
wget -P data/ https://s3.amazonaws.com/pkerp/public/$HITILE

python manage.py ingest_tileset --filename data/$COOLER --filetype cooler --datatype matrix --uid cooler-demo
python manage.py ingest_tileset --filename data/$HITILE --filetype hitile --datatype vector --uid hitile-demo

We can now use the API to get information about a tileset, or to get the tile data itself:

curl http://localhost:8000/api/v1/tileset_info/?d=hitile-demo
curl http://localhost:8000/api/v1/tiles/?d=hitile-demo.0.0.0

Development

Start the server:

python manage.py runserver localhost:8001
// or
npm start

Test the server:

./test.sh
// or
npm test

Bump version of server:

bumpversion patch

Update source code:

./update.sh

Troubleshooting

pybbi installation fails on macOS: Check out https://github.com/nvictus/pybbi/issues/2

macOS 10.15 dependencies

• brew install hdf5
• brew install libpng
• brew install jpeg
• FUSE for Mac

License

The code in this repository is provided under the MIT License.