Migrating Documentation from WIKI to README

[sappelhoff]

This PR fixes #74, by making the documentation on available datasets more accessible and providing a table that shows (approximately) which modalities are present in each dataset.

Also fixes #154 to some extend, by adding all available links to the full data. The rest will have to be added by the respective dataset-authors later on.

The table is quite large although I tried hard to cut down on complexity. I would be very grateful for a review and some feedback.

Note that I also inserted a comment in the README, that we are interested in new examples as long as ... they cover some aspects that are not already covered.

I don't see a point of hosting many datasets here if they are all the same in terms of their structure. Personally, I would even be up for cutting down on some example datasets we are hosting here - so please let me know about your opinion on this as well.

• Upon merge, delete Wiki

---

Rendered version: https://github.com/sappelhoff/bids-examples/blob/overhaul_docs/README.md

Google Sheet version: https://docs.google.com/spreadsheets/d/1d9KfmaKvTOpFWgi0Kd-tZGgVTsZk_4BHH9prcgFeigw/edit?usp=sharing

[effigies]

LGTM. Very small comments.

[choldgraf]

The formatting looks good to me - I agree that the signal-to-noise is hurt by having so many datasets with overlapping structure. I'd rather something like:

1. Figure out the different patterns that should be highlighted, make sure there's one dataset that does this well.
2. Add links to other examples of the same dataset, or maintain a separate page that simply lists all of the datasets that are accessible (maybe just link to OpenNeuro?).

The challenge with the current setup is that the repo serves two purposes: one is to show what the BIDS structure is like, another is to list lots of different examples. I think these are two separate goals that deserve two separate pages. (though doesn't need to be in this PR)

[franklin-feingold]

thank you @sappelhoff for raising this! I can help fill in the bit.ly for openneuro sets. We tried to preserve the ds assession number consistency when merging from openfmri to openneuro. I agree with @choldgraf for two different lists. The more categorized by type (with a coupe of representative examples) can be the more top level/easy to find and if a user needs more examples within a category then this can be linked out to the master list with the collection of all the datasets. The category classification may be challenging to determine the optimal level of abstraction. I think best in a different PR.

[sappelhoff]

Thanks for the round of reviews everybody!

Let's move on and get this PR ready for merging and then tackle the next issue of "making the overview more concise" in a new PR.

@franklin-feingold I'll give you edit access to the Google Sheet, and it'd be great if you could help me filling it up with the missing information / links to openneuro (instead of the bit.ly ones)

One additional thing: For the modality columns in the table, I'll change the x markings of when a modality is present to a repetition of that modalities name (mri, eeg, ...) --> this will make it easier to scroll down a long table and still know for each row, which modalities are present, without having to go back up to the header.