Taylor Salo

I believe that it would look something like this [completely untested code]: ```python def computefeats2(data, mmix, mask, n_iters=1000): """ Converts `data` to component space using `mmix` Parameters ---------- data :...

@smoia You've been digging into MELODIC's code a bit. Do you know how they convert their component maps into z-statistics?

If we just use more aggressive dimensionality reduction in the PCA step _prior_ to metric calculation, then the properly calculated statistics should be valid. We can do this by setting...

If we want to go with the permutation approach, it looks like [nilearn.mass_univariate.permuted_ols](http://nilearn.github.io/modules/generated/nilearn.mass_univariate.permuted_ols.html#nilearn.mass_univariate.permuted_ols) does what I was thinking, so it's an issue that's been dealt with already I guess. It...

This issue should be somewhat resolved by our shift to MA-PCA methods (see https://github.com/ME-ICA/tedana/issues/178#issuecomment-512807879), and we decided to refactor the regression function instead of using a nonparametric approach like `permuted_ols`....

This is coming up as well in the BrainHack Donostia project [`ica-aroma-reorg`](https://github.com/Brainhack-Donostia/ica-aroma-org), where we need a solid version of `computefeats2` to reproduce MELODIC's outputs.

I think that the current solution is available [here](https://github.com/CesarCaballeroGaudes/tedana/blob/c3b5ece14b1431c78675fcd29002cb09e3c34cc3/tedana/stats.py). @CesarCaballeroGaudes @smoia do you know what, if anything, still needs to be done to get it working? EDIT: I'm happy to...

That would be amazing. Thank you!

This is something we have been discussing in #820, but the problem is that the CFF format doesn't currently support multiple citations very well, which is a problem when we...

I am still concerned for a few reasons: 1. Even though your CFF file has multiple references, even selecting the BibTeX option in "Cite this repository" doesn't export those references....