What type of tool fits a "Federated framework"?

[matuskalas]

From @carleshf, thank you for the request🙏🏽

We discussed internally how we want to proceed and decided that we will
ask out collaborators to register the tools they want to use in EUCAIM
project in bio.tools beforehand. The only issue we see is the lack of a
label related to "federated analysis". on one side we see that an
analytical tool can be registered as statistics for instance, but there
is no way a "federated framework" can be registered with the current
wording. Would it be possible to add "federated framework" as a label in
bio.tools ontology?

The questions I have in mind are:

• Which of the current tool types fits this the best, if any?
• Do we need to add something like that?
• If we add it, what else should we also add? Various kinds of "Distributed framework"/"Distributed resource"/..., or just one?
• What should it/they be called, to keep it as "standard" as possible?

[veitveit]

One could keep it simple and annotate the tool type of the software accessing and using the distributed framework.
This could then be anything from GUI, web interface to library.

What is the difference between a federated framework and a platform that uses different resources by e.g. accessing their APIs for running its own analysis?

I would see the annotation more on the EDAM side (operations and topic).

[matuskalas]

One could keep it simple and annotate the tool type of the software accessing and using the distributed framework. This could then be anything from GUI, web interface to library.

What is the difference between a federated framework and a platform that uses different resources by e.g. accessing their APIs for running its own analysis?

I think this point and question make a lot of sense👍🏽

I would see the annotation more on the EDAM side (operations and topic).

I would like to avoid modelling such aspects in EDAM, unless it is somewhat scientifically justified (i.e. not purely technical)

[carleshf]

Let me try to explain my point of view.

One could keep it simple and annotate the tool type of the software accessing and using the distributed framework.
This could then be anything from GUI, web interface to library.

Distributed is a general concept different from federated. Here I am talking about data availability. In a "distributed environment" one assumes the availability of the data. In a "federated environment", the data is exclusive to each node. Is a tool that works for federated analysis extensible to a distributed environment? The answer is, depends on the tool.

What is the difference between a federated framework and a platform that uses different resources by e.g. accessing their APIs for running its own analysis?

Any tool that works in a federated way should (special emphasis on the "should" here) take some sort of care on how the data is managed and accessed, and which and how results are reported.

I would see the annotation more on the EDAM side (operations and topic).

I am not the one to discuss this point. I have full trust in you @matuskalas

---

The questions I have in mind are:

• Which of the current tool types fits this the best, if any?

Within the project we are talking, EUCAIM, we expect to see some frameworks that will fit under that category, as well as some tools that only make sense under that typology (aka, model aggregators).

• Do we need to add something like that? If we add it, what else should we also add? Various kinds of "Distributed framework"/"Distributed resource"/..., or just one?

As I understand, it should not be necessary to add the full set of tags if we can select them separately. For instance, "distributed" vs "federated" sounds good, but "framework" is, in the end, a tool that could fall under "Workbench" (that already exists). But, as @matuskalas highlights, "Resource" is also a really good candidate for "databases-like" that "do nothing" but provide data.

• What should it/they be called, to keep it as "standard" as possible?

Following some examples, like the "Portal" issue, I was proposing just "Federated" so we could pair it with other labels from EDAM to make sense of it. If "Distributed" and "Resource" are added too, with "Workbench", I think we are covering the spectrum. I'm going to ask some colleagues to come to the thread and see what they say.