Gen3 - Working with the API

Using the API

What Does the API Do?

The API is created programmatically based on the Gen3 commons data model. All of the work Gen3 data contributors do to prepare their metadata powers the API (see steps 4-6 in the Data Contribution section).

With the API in place, users can submit queries to find metadata information across the Gen3 commons. The API can be queried programmatically or through provided tools, like the submission portal.

The Gen3 commons uses GraphQL to manage the metadata. To learn the basics of writing queries in GraphQL, please visit:

Credentials to Query the API

The credentials that allow access to “raw” data in the object store and ssh keys to access VMs, also allow users to programmatically query or submit data to the API. This credential is used every time an API call is made.

Each API key is valid for a month and is used to receive a temporary access token that is valid for only 30 minutes. The access token is what must be sent to the Gen3 API to access data in the commons.

For users granted data access, the API key is provided on the Profile page after clicking the “Create API key” button.

Profile Page

While displayed, click “copy” to copy the API key to the clipboard or “download” to download a “credentials.json” file containing the id/key pair in json format.

Copy Key

In python, the following command is sent, using the module “requests”, to receive the access token:

# Save the copied credentials.json from the website into a variable "key":
key =  { "api_key": "<actual-key>", "key_id": "<a-key-uuid>" }

# Import the "requests" python module:
import requests

# Pass the API key to the Gen3 API using "" to receive the access token:
token ='', json=key).json()

# Now the access_token should be displayed when the following line is entered:

When submitting a graphQL query to the Gen3 API, or requesting data download/upload, include the access token in the request header:

headers = {'Authorization': 'bearer '+ token['access_token']}

# A GraphQL Endpoint Query Using the "key" JSON:
query = {'query':"""{project(first:0){project_id id}}"""};
ql ='', json=query, headers=headers)
print(ql.text) # display the response

# Data Download via API Endpoint Request:
durl = '<program>/<project>/export?format=tsv&ids=' + ids[0:-1] # define the download url with the UUIDs of the records to download in "ids" list
dl = requests.get(durl, headers=headers)
print(dl.text) # display response

# Data Upload via API Endpoint Request:
headers['content-type']='text/tab-separated-values' # add the content-type to header
u = requests.put('', data=tsv, headers=headers)

If an an error such as “You don’t have access… ” occurs, then the API key is most likely out of date and a new access token will need to be made.

Queries in the Submission Portal: GraphQL

Queries can directly run in the submission portal by clicking the “Query” magnifying glass or directly at: The query portal has been optimized to autocomplete fields based on content, increase speed and responsiveness, and generally make it easier for Gen3 members to find information.

NOTE: For these user guides, is an example URL and will not be the actual URL of the data commons.

GraphQL Query

Pagination and Offsets

Queries by defult return the first 10 entries. To return more entries, the query call can specify a larger number such as (first:100).

In the case that too many results are returned, a timeout error might occur. In that case, use pagination to break up the query.

For example, if there are 2,550 records returned, and the graphiQL query is timing out with (first:3000), then break the query into multiple queries with offsets:

(first:1000, offset:0) 		# this will return records 0-1000
(first:1000, offset:1000) 	# this will return records 1000-2000
(first:1000, offset:2000) 	# this will return records 2000-2,550

Updating the example template details from experiment sample query to call the first 1000, the call becomes:

	"query":" query Test {
		experiment (first:1000, submitter_id: "<INSERT submitter_id>") {  
	} "

Browsing by Project Node

The metadata submission portal can be used to browse an individual submission by node. Select a project and then click the “browse nodes” button to the right. From this screen queries can be made by node in the dropdown at the left.

Example: Browse by Node

Browse by node

This feature can also download the tsv associated with the node, or if a user has “write” access to the this project, delete existing nodes.

Graphing a Project

A user can review a graph of an individual project, by toggling between views of the completed nodes and the full graph.

Example: Graphing a Project

Graphing a project

Using the Gen3 SDK

The bioinformatics team at the Center for Translational Data Science (CTDS) at University of Chicago has put together a basic software development kit (SDK) to help users interact with the Gen3 API, which can be found on Github. The Gen3 community is encouraged to add to the functions library and/or improve the notebook.

NOTE: As the Gen3 community updates repositories, keep them up to date using git pull origin master.

To install the Gen3 SDK, use the python installer ‘pip’.


# Install Gen3 SDK:
pip install gen3

# To clone and develop the source:
git clone
Back to Managing Timepoints