Any roadmap for fitlins?

[yarikoptic]

Hi Everyone,

Now that Derivatives (BEP003) and Linear Models (BEP002) are getting into somewhat usable shape, I wonder what are the plans/schedule for the fitlins -- will it be developed soon to accommodate for all the changes in those BEPs or there is another module/framework in works to look at in attempt to make initial use of BEP002?

Cheers!

[effigies]

Hi Yarik,

We're just about finished with getting fMRIPrep Derivatives RC-compliant, and so the next thing to turn to is FitLins. I hope to smooth out any derivatives-related issues this week, and be working on models by next week.

[yarikoptic]

Awesome! Thank you @effigies !

[yarikoptic]

I guess the majority of the work is to be done within https://github.com/bids-standard/pybids/tree/master/bids/analysis ?

[effigies]

Generally, yeah. And in https://github.com/poldracklab/fitlins/blob/master/fitlins/interfaces/bids.py

[fliem]

Hi @effigies
is there a timeline for the next release?
Is there an example model-file that replicates SPM first level results and that works with the current master?
Thanks!

[effigies]

Hi @fliem. I just looked at the merge log, and it looks like I should definitely cut a release ASAP. This week is super busy, so I can't promise anything before next week, but if you're comfortable installing from master, then please feel free.

For model files, I will check. Do you have an open dataset you're looking at, or would you be happy with any OpenNeuro dataset?

[fliem]

Thanks @effigies.
I installed from master, but for using it for a real analysis, I'd prefer a released version (is there anything speaking against using fitlins for real projects at the moment?).
Any dataset would be fine. It seems that the model examples are outdated, but I have too little insight into the BEP002 development process.
Always happy to contribute if there is anything I can do (for instance examples or documentation...).

[effigies]

The main problems for real analyses right now are:

1. Second-level analyses (i.e., anything above run-level) do not take into account variance estimates, only beta contrasts.
2. Poor reporting at the second-level.
3. No F-contrasts.

And yes, contributions are always welcome. Which I guess a roadmap would be useful for. In particular, tests and documentation are basically missing, but what would interest you to contribute most?

Is bids-standard/bids-examples#130 where you've been looking for examples? If so, I'll prioritize rewriting those to the current specification today. BEP-002 is a pretty slow-moving target at the moment, but it's also not fully internally consistent, so I understand that it's pretty confusing. That might be a good flight activity for me, tomorrow...

[fliem]

For our purpose, run-level analysis is all we need.
I could contribute to the documentation.
I didn't realize this PR existed, I only have looked at fitlin's models branch.
It would be great to validate a very simple run-level contrast created with fitlins agains the same contrast created with something established (like SPM).