Merge Bep010 ieeg examples onto master

[sappelhoff]

closes #131
Apart from rebasing on master, this PR brings two changes:

1. bump up travis version: 49cc200
2. zero out two large datafiles in ieeg_visual: cfac0a4

[chrisgorgo]

LGTM!

[sappelhoff]

this is relevant to: https://github.com/bids-standard/bids-validator/issues/749

I'd like to merge these examples soon and then make a release of the bids-examples to be tested with the bids-validator.

One more approving review from @dorahermes or @choldgraf should do the trick :-)

[dorahermes]

Just a couple of small suggestions for the epilepsy dataset and I found a small error in one of my datasets for the electrode size. I will make a pull request to fix my dataset.

[sappelhoff]

@dorahermes do you approve of merging this PR now?

[dorahermes]

The only comment left is that the electrode names in the ieeg_epilepsy dataset contain a non-letter/number symbol: '. I cannot fix this according to the data/header, because they have zero bites and I can't read the channel labels. I could just remove the ', or just leave it since the dataset currently passes the validator? @choldgraf

[sappelhoff]

What was the reason to only allow numbers and letters in the first place? We don't have this restriction in EEG, and apparently we missed this inconsistency between EEG and iEEG 😬

[choldgraf]

arggg - what's the state of the validator? I'm guessing it'll pass both of them even if there's a symbol in the channel names, is that right?

[sappelhoff]

I'm guessing it'll pass both of them even if there's a symbol in the channel names, is that right?

yes, that's right :-)

why do you think that symbols in the channel name are a bad idea?

[dorahermes]

The initial thought was that it would be more difficult to match electrode names and channels labels in case of symbols.

[sappelhoff]

The initial thought was that it would be more difficult to match electrode names and channels labels in case of symbols.

Do you have an example where it would be more difficult? Just so that I can understand better?

To move further, I see these options:

1. we change the spec to also allow symbols ... this would not be backward compatibility breaking
2. we enforce to not allow symbols. This would mean:
   • asking @ftadel to adjust his dataset so that channel/electrode names do not contain symbols
   • implementing a check in the validator that prevents symbols in channel/electrode names

[choldgraf]

if symbols can be in both the channels and in the electrodes names, then I'd be +1 on supporting them since those symbols would be "just another character in a string".

e.g., If channels and electrodes were named ch-01 I could see that being easier to parse then ch01

[choldgraf]

(but, this would require a change to the spec, as @sappelhoff says...so to me the question breaks into "do we change the spec, or do we change the validator")

[sappelhoff]

+1 for adding to the spec that symbols are also permissible as this would also increase consistency with EEG and I don't see a convincing reason to not include symbols

[dorahermes]

+1 for adding to the spec that symbols are also permissible for the same reasons

[ftadel]

+1 for allowing symbols
Many epilepsy centers use a prime (') to indicate whether an electrode is implanted in the left or the right hemisphere (eg. leading to channels A01..A12 and A'01..A'12). Not allowing the single quotes and some other variations (commas, dots, ...) would make the formatting into BIDS more complicated.

[sappelhoff]

alright! see bids-standard/bids-specification#182 ... please review :-)

[sappelhoff]

just fyi: Unless something happens, I will merge this one once bids-standard/bids-specification#182 is merged

[choldgraf]

I think @dorahermes should "un-request changes" before we merge, just to respect the community process...

[sappelhoff]

I think @dorahermes should "un-request changes" before we merge, just to respect the community process...

+1

[choldgraf]

🎉