Vantage6

This repository is contains all the vantage6 infrastructure source code. The vantage6 technology enables to manage and deploy privacy enhancing technologies like Federated Learning (FL) and Multi-Party Computation (MPC). Please visit our website to learn more!

You can find more (user) documentation at readthedocs. If you have any questions, suggestions or just want to chat about federated learning: join our Discord) channel.

Infrastructure overview

A High level overview of the vantage6 infrastructure. Vantage6 has both a client-server and peer-to-peer architecture. The client is used by the researcher to create (PET) computation requests. It is also used to manage users, organizations and collaborations. The server contains users, organizations, collaborations, tasks and their results. It provides a central access point for both the clients and nodes. The nodes have access to privacy sensitive data and handle computation requests retrieved from the server. Computation request are executed as separate containers on the node. These containers are connected to containers at other nodes by a VPN network.

:books: Quickstart

Requirements

The vantage6 infrastructure is delivered in Docker images. To run these images, you need to have Docker installed. To install the latest version of the vantage6 CLI, you need to have Python, we recommend using an environment manager like mini-conda.

Install the latest version of the vantage6 CLI by using:

pip install vantage6

This install the v6 commands, which allows you to manage your nodes and servers. To view all available options, run:

v6 --help

For example you can create a local test setup by using:

v6 dev create-demo-network

This creates a local network with a server and two nodes. You can start the network by running:

v6 dev start-demo-network

This will start the server and nodes in the background. You can view the logs by running:

# View node logs
v6 node attach

# View server logs
v6 server attach

From here you can use the vantage6-client to interact with the server. The demo network has a pre-configured organization with the following credentials:

• Username: dev_admin
• Password: password

For example, you can create a new organization by running:

from vantage6.client import Client

client = Client('http://127.0.0.1', 7601, '/api', log_level='debug')
client.authenticate('dev_admin', 'password')
client.setup_encryption(None)

client.organization.create(
    name='My organization',
    address1='My address',
    address2='My address',
    zipcode='1234AB',
    country='The Netherlands',
    domain='my-organization.com'
)

You can find more (user) documentation at readthedocs

Project structure

PYPI packages

This repository is home to 6 PyPi packages:

• vantage6 -> CLI for managing node and server instances
• vantage6-client -> Python client for interacting with the vantage6-server
• vantage6-algorithm-tools -> Python tools to facilitate algorithm development
• vantage6-node -> Node application package
• vantage6-server -> Server application package
• vantage6-algorithm-store -> Algorithm store application package
• vantage6-common -> Package with common vantage6 functions
• vantage6-backend-common -> Package with functions common to central server and algorithm store

Note that when using vantage6 you do not install the server and node packages. These are delivered to you in Docker images.

This repository also hosts the code for the vantage6 user interface (UI). The UI is an Angular web application that can be used to interact with the vantage6 server easily.

Docker images

The vantage6 infrastructure is delivered in Docker images. All Docker images are stored in our private Harbor registry. The most important images are:

• harbor2.vantage6.ai/infrastructure/node:VERSION -> Node application Docker image
• harbor2.vantage6.ai/infrastructure/server:VERSION -> Server application Docker image
• harbor2.vantage6.ai/infrastructure/ui:VERSION -> User interface Docker image
• harbor2.vantage6.ai/infrastructure/algorithm-store:VERSION -> Algorithm store Docker image

with VERSION being the full semantic version of the vantage6 infrastructure, e.g. 4.0.0 or 4.1.0rc0.

Several other images are used to support the infrastructure:

• harbor2.vantage6.ai/infrastructure/infrastructure-base:VERSION -> Base image for the infrastructure
• harbor2.vantage6.ai/infrastructure/squid:VERSION -> Squid proxy image used for the whitelisting service
• harbor2.vantage6.ai/infrastructure/alpine -> Alpine image used for vpn traffic forwarding
• harbor2.vantage6.ai/infrastructure/vpn-client -> VPN image used to connect to the VPN
• harbor2.vantage6.ai/infrastructure/vpn-configurator -> VPN image used for initialization
• harbor2.vantage6.ai/infrastructure/ssh-tunnel -> SSH tunnel image used for connecting algorithms to external services

And finally there are some images released for algorithm development:

• harbor2.vantage6.ai/infrastructure/algorithm-base:MAJOR.MINOR -> Base image for algorithm development
• harbor2.vantage6.ai/infrastructure/algorithm-ohdsi-base:MAJOR.MINOR -> Extended algorithm base image for OHDSI algorithm development

:gift_heart: Join the community!

We hope to continue developing, improving, and supporting vantage6 with the help of the federated learning community. If you are interested in contributing, first of all, thank you! Second, please take a look at our contributing guidelines and our code of conduct.

:scroll: License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.

:black_nib: Code of Conduct

Please note that this project is released with a Contributor Code of Conduct. By participating in any way in this project you agree to abide by its terms.

:black_nib: References

If you are using vantage6, please cite this repository as well as the accompanying papers as follows:

• F. Martin, M. Sieswerda, H. Alradhi, et al. vantage6. Available at https://doi.org/10.5281/zenodo.7221216. Accessed on MONTH, 20XX.
• A. Moncada-Torres, F. Martin, M. Sieswerda, J. van Soest, G. Gelijnse. VANTAGE6: an open source priVAcy preserviNg federaTed leArninG infrastructurE for Secure Insight eXchange. AMIA Annual Symposium Proceedings, 2020, p. 870-877. [BibTeX, PDF]
• D. Smits*, B. van Beusekom*, F. Martin, L. Veen, G. Geleijnse, A. Moncada-Torres, An Improved Infrastructure for Privacy-Preserving Analysis of Patient Data, Proceedings of the International Conference of Informatics, Management, and Technology in Healthcare (ICIMTH), vol. 25, 2022, p. 144-147. [BibTeX, PDF]