Skip to content


Repository files navigation


Manage HiGlass Instances

This utility script helps manage local HiGlass instances


To use this utility, you will require:


higlass-manage can be installed using pip

pip install higlass-manage


HiGlass wraps the Docker commands for starting, stopping, listing and populating local higlass instances.


To run the tests, first get the test data:


And then run the tests:


Quickly viewing a dataset

The simplest way to get started is to open and view a dataset. The higlass-manage view command will automatically start a new instance if one isn’t already running, add the given dataset and display it in a browser. Currently, the higlass-manage view command only works with cooler, bigWig, chromsizes and gene-annotation files.

higlass-manage view

Starting a HiGlass instance

Start a local higlass instance using the default data and temporary directories: ~/hg-data and /tmp/higlass-docker. All of the data ingested into the instance will be placed into the data directory. Alternate data and temp directory can be specified using --data-dir and --temp-dir parameters.

higlass-manage start

If you want to make your instance accessible to the outside world, you need to specify the host URL that it will be available through using the --site-url parameter:

higlass-manage start --site-url

These commands will start an instance running on the default port of 8989. An alternate port can be specified using the --port parameter. The number of worker processes for the uWSGI application server can be specified with the --workers parameter.

Using the Redis caching service

To make use of the Redis caching service to improve performance, add the --use-redis flag. Redis files will be stored by default in the ~/redis-data directory. Add the --redis-dir parameter to override this default.

higlass-manage start ... --use-redis --redis-dir /new/path/to/redis-data

Setting default client options

To the default options for newly created tracks, use the --default-track-options parameter to pass in a JSON file containing either track-specific or general default track options:

$cat default_options.json
    "all": {
        "showTooltip": "true"

$ ./ start --default-track-options default_options.json

Ingesting data

Use the ingest command to add new data. Generally data requires a filetype and a datatype. This can sometimes (i.e. in the case of cooler and bigwig files) be inferred from the file itself.

higlass-manage ingest my_data.mcool

In other, more ambiguous cases, it needs to be explicitly specified:

higlass-manage ingest my_file.bed --filetype bedfile --datatype bedlike --assembly hg19

Note that bedfiles don't store chromosome sizes so they need to be passed in using either the --assembly or --chromsizes-filename parameters.

Listing available datasets

pete@twok:~/projects/higlass-manage$ higlass-manage list tilesets
VlWKy6ofT6qMFGf-uG_5pQ | beddb | bedlike | GSE93955_CHIP_DMC1_B6_peaks.bed.multires
LAXFhHhASa2zDgJRRS67cw | cooler | matrix |

Starting a shell

For debugging purposes it can be useful to run a shell within the Docker container hosting the higlass instance. This can accomplished using the shell command:

higlass-manage shell

Getting the error logs

When errors occur they are usually on the higlass-server end. To output the logs use the logs command:

higlass-manage logs

Stopping a HiGlass instance

To stop a running instance, use the stop command:

higlass-manage stop

Migrating a HiGlass instance

Migrating a higlass instance between different servers can be done by copying the data-folder, typically hg-data, from server of origin to the destination and re-starting higlass:

# at the destination:
scp -r  /new/path/
higlass-manage start --data-dir /new/path/hg-data ...

Tilesets ingested at the origin would be available at the destination. However, viewconf-s saved at the origin would not work at the destination, because the tilesets would be referred there with original URLs, e.g.

This can be fixed by updating viewconfs in the database before copying hg-data:

# at the
higlass-manage update-viewconfs --hg-name-old old_hg_name --new-site-url

in this case, higlass instance old_hg_name would be used to infer old site URL, port and path to the data folder.

Same can be achieved even without any running higlass instances, but then one has to provide path to the data folder and site's URL and port both "new" and "old":

# at the
higlass-manage update-viewconfs --old-site-url --data-dir /old/path/to/data --new-site-url

update-viewconfs would save updated database as /old/path/to/data/db.sqlite3.updated and keep the original /old/path/to/data/db.sqlite3 unchanged. Thus, db.sqlite3.updated has to be renamed to db.sqlite3 after migrating to


The following is a list of handy commands when developing HiGlass:

  • Start locally built docker image: The locally built image must be named image-default. Usually built using higlass-docker.
    higlass-manage start --version local


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