Could you introduce yourself?

I am a physicist by training. After my PhD (in 2006) I joined a cell biology lab to work on a microscope prototype. I also started developing image analysis and modelling software within the same lab.

I then worked for some time as an independent consultant, before joining the CENTURI multi engineer platform in Marseille, and France-BioImaging two years later in 2022.

I tried early on to deploy OMERO in my lab (in 2009, I did not succeed), I am convinced good data management is both a good practice and a huge time saver for research teams in the long run.

At France BioImaging, I am a research software engineer in the FBI.data team. As the senior engineer in the team, I overview software design and the exchanges with our national and international partners.

You are strongly involved in FBI.data project, what are the main goals of this team?

The main focus of the project is the development and deployment of the BioImage Cloud solution.

The goal is to provide the French research community with tools and standards to ease microscopy data management, federate computation and storage resources and help the adoption of FAIR (Findable, Accessible, Interoperable and Reusable) practices.

What are the origins of BioImaging Cloud? How did this idea first emerge?

The project developed as part of the ANR “equipex” project Mudis4LS. The overall objective of this project, a collaboration between the French Bioinformatics Institute (IFB) and FBI, is to advance the French Government Open Science roadmap.

In practice, for bioimages, the stated goals imply a lot of technical and human constraints:

• we have to manage data with huge disparities in shape and size, that are produced in facilities siloed inside a local network, with often restricted availability of IT staff.
• furthermore, FAIR data culture is still developing in the field, so we also need to offer easy to use tools, as a way to favor the change of mindset toward an open science / FAIR culture.

So we started by stating these constraints explicitly, and developed the tool that would satisfy them.

In simple terms, how does BioImage Cloud work?

BioImage Cloud works at two levels. The lower one is the network layer between a facility and a mesocentre, the regional datacenter hosting the service. BioImage Cloud relies on setting up a ‘tunnel’ out of the facility’s private network toward the mesocenter. The strength of the solution is that this is done securely and reduces the IT work load involved with data transport. Of course users need not concern themselves with this layer, but it’s actually quite important that it is there.

The second level is an ensemble of software tools that provide users with a way to annotate, transport and share their data. In not too much detail, once properly described, the data are copied into the federated storage and imported into OMERO, the standard tool for microscopy data management. From there, they can be shared with collaborators, collaboratively annotated, and (soon) analysed on the mesocenter high performance computing resources.

What are the main benefits for the users?

First and foremost it is easier to access all the microscopy data created in your project, with more entry points, and a standard organization and annotation.

Concretely, this means that a PhD student can readily share their latest experiment results with their advisor, and the latter can access those results from anywhere, including once the thesis is over and the student is gone.

For the experimenter, rich annotations means it is easy to filter through experimental conditions, or write the material and methods section.

It also makes life much easier for bioimage analysts, as relevant metadata (e.g. the scale of the image) is readily available in the system, and API access allows to seamlessly shift from a prototype to batch processing of large datasets.

From a practical point of view, what does a user need to do to use BioImage Cloud?

First you need to belong to a French research institution, to create an account on the platform. To import data in the solution, you need access to a collection server, provided by the microscopy facility you work with.

With that set up, data import works in two steps: you first need to upload the data themselves through FTP on the collection server and second trigger the import in OMERO by uploading a spreadsheet describing the data.

Creating this spreadsheet is the most time intensive part of the process, as we ask our users for a detailed description of the experiment behind the data. Yet, the spreadsheet can be if you repeat the experiment, are slightly altered to account for new experimental conditions. So the process is faster and easier after the first time you or someone on your team use the service.

On which facilities is BioImage Cloud already available?

As of today, 8 servers are deployed in 7 cities. For the south-east part, BioImage Cloud is available in Montpellier for all MRI (CRBM + IGH), Marseille in Luminy, Strasbourg at IGBMC, Lyon at LyMic and for the south-west Nantes (at IRS), Rennes at INRIA and Paris at IBENS.

Two additional sites are under deployment, Grenoble and Toulouse, and 5 more servers will be deployed before the end of 2026.

What are the next steps for BioImage Cloud?

There are many new features down the road. The ones I’m most excited about are:

• the support of the OME-ZARR format, adapted to big data, for example from light sheet or spatial omics experiments;
• the automated submission of published data to BioImage Archive (in collaboration with the IFB madbot team);
• the easy access to high performance computing for bioimage analysts, e.g. for heavy deep learning workflows.

What message would you like to share with imaging facility staff or researchers who may be interested in using BioImage Cloud?

First, that they are very welcome and we are here to help understand their use cases!

We hold open desk sessions every Monday and Wednesday afternoon, so the door is open. A big goal for us this year is to adapt the solution to their usage, so we are eager for feedback.

Second, the tool is new so there are bugs, so I’d ask for their patience! We hope we’ll have more and more users this year that will help consolidate and shape a nice tool.

If you jump in early, you’ll be able to brag in a few years that you were there in the heroic era, to not miss that opportunity!

If you are interested in BioImage Cloud, you can find useful links at the end of our dedicated article, including access to the open desk sessions and the mailing list.

You can also watch the replay of the FBI Connect webinar, where Guillaume Gay presented BioImage Cloud and its main features.