Taylor Salo

Here's what the `minimal_nativeres` bunch looks like: ``` {'dataset_description.json': ['/Users/taylor/tedana_data/cambridge/dataset_description.json'], 'description': 'Notes\n' '-----\n' 'The original data is downloaded from OpenNeuro\n' 'https://openneuro.org/datasets/ds000258/versions/1.0.0\n' 'This fetcher downloads preprocessed data that are available '...

I like the idea of suppressing the warnings. I assume we'll need to bump our minimum versions for several packages once numpy removes these aliases?

For the most part, negative weights don't seem to have much structure, but in the global signal components both positive and negative weights seem to follow patterns. It's really hard...

If the goal of component selection is to identify which components are S0-based and which are R2-based, and one of the metrics we use to separate noise from signal is...

@smoia The 20 voxel threshold is really [`np.max([int(n_voxels * 0.0005) + 5, 20])`](https://github.com/ME-ICA/tedana/blob/f9ffca1b26e77e94aeda7c6945932c8606524343/tedana/metrics/kundu_fit.py#L237), but in practice (on our test datasets), that ends up being 20. That threshold should increase with...

We flip components separately. Since the components are initially arbitrarily signed, some might need to be flipped, while others might not, to best match the data.

It's performed [here](https://github.com/me-ica/tedana/blob/65f89e1/tedana/model/fit.py#L108-L113): https://github.com/ME-ICA/tedana/blob/65f89e1629bfddcf7f01a91ccb052440a6c22ea4/tedana/model/fit.py#L108-L113

Since component flipping doesn't pose a problem for reconstruction, how do we feel about negatively weighted voxels? Is the consensus that it's okay to treat those voxels equivalently to positively...

I agree with @jbteves' points for the most part. I think that the burden of reorganizing and summarizing the changes at the time of a given release are lower than...

@dowdlelt Sorry, somehow I never saw your response. It does look like some voxels in CSF are flagged with only one good echo, but I think that's okay. I haven't...