[BUG] Undo not working after addition of annotation building block.

[Brilator]

Describe the bug
Excel "Undo" is not working properly after addition of an annotation block.

To Reproduce
Steps to reproduce the behavior:

1. Go to 'Annotation building block selection'
2. Search for any term to add as annotation column.
3. Click on 'Annotation building block'
4. Try to undo (ctrl+z or edit-> undo)
5. See interesting behavior, which I cannot properly put into words. It at least does not simply delete the annotation block, which is probably due to multiple columns being added in parallel (including the hidden ones).

Expected behavior
Undo should undo the complete last action of annotation block addition.

OS and framework information (please complete the following information):

• OS: macOS Catalina
• OS Version 10.15.7
• MS Excel: Excel Desktop
• MS Excel Version 16.43

Additional context
I must admit that I also find the hidden cols a bit annoying. I understand the rationale to not overwhelm standard users with ontologies. But I'd rather try to teach users why ontologies matter than hiding data in the background.

[Brilator]

Ah, sorry. I think this is kind of related to #48.

[Freymaurer]

This is actually related to the issue mentioned here: #9 .
We are currently not able to track changes done by the add-in and i think it even deletes all changes it tracked to that point. I am not sure if we even can fix that issue without implementing a version history on our own.

@Brilator Your feedback is very welcome! Thank you for all your input.

[kMutagene]

Regarding undo functionality, this is disabled by-design of the underlying office.js library (and at that, specific to excel). Not much we can do about that.

I agree that there could/should be an option to disable the auto-hiding though.

The problem i see with the general format is that the ontology references do not add much to the flow of reading such an annotation file (from the 'what happened from source to sink' perspective). Ontologies matter, but i am not sure if a column with 20 times the same url does really help at that. I think making a lot more settings optional/togglable is the way to go here, and maybe enable saving settings down the line so you don't have to unhide those columns every time if you are interested in them. At the end we want to enable both styles of working here, more in-depth focus on the annotations as well as straight-forward workflow annotation

[Brilator]

Hey, no worries, there will be more input...
You're doing a great job with ARC, SWATE and annotation principles!
Not sure if Timo explained, I'm Dominik from Düsseldorf (HHU, CEPLAS). GitHub Newbie btw, so don't hesitate to let me know if I have to do things differently (e.g. issue templates are helpful).

[Brilator]

One more thing here, since I'm not sure wether this is a bug or added feature:
Some characteristics that may be specific to a lab or facility (e.g. metabolomics) will probably never have a "Term Source REF" and "Term Accession Number" (for example the "LabelText" in your anno principles slide). Is it necessary to still automatically add these to stay ISA compliant?

[kMutagene]

@Brilator Timo mentioned that there will be feedback coming in from you. Don't worry, your input is very valuable.

Some characteristics that may be specific to a lab or facility (e.g. metabolomics) will probably never have a "Term Source REF" and "Term Accession Number" (for example the "LabelText" in your anno principles slide). Is it necessary to still automatically add these to stay ISA compliant?

There are two answers for that. First, you can annotate a characteristic with free text, without using an ontology term. See here. The accession number and term source ref fields are only required when there is an ontology term in use. When such a term does not exist, these columns can stay empty.

The preferred approach to this would however be filing a missing term issue at our ontology so we can add the missing term to our database (which will also lead to the terms being available in the term search). That way, we can ensure to both fill the ontology gap and make sure that workflows are as well annotated as possible.

[Brilator]

Yes sure, I know you can add free text and leave cells blank for ontology-unrelated terms. But this answered my question - it's required for ISA compliance.
And sure in the long-term we'll want to have everything referenced via ontology / PIDs, but there will always be user/lab-specific processing steps (e.g. storage location of samples / extracts: the 20 °C freezer in basement room 4 - board 6 - box 2). These will not matter for downstream analyses, but I would still want to include for complete traceability and to push user-friendliness (i.e. everything in one place).

[kMutagene]

Sorry i misunderstood the question. When there is no intent to add a term later for such a characteristic, the empty columns are not required, as ontology annotations themselves are optional in the ISA format.

[Brilator]

Ah, alright. Thanks!