Add project, data collection metadata and associated CV tables

pbishwakarma commented 1 year ago

Adding back in project and data collection for review according to issue process.

Closes #21

patrick-lloyd-ray commented 11 months ago

@pbishwakarma and @carolth: it's in standard MOWG form now but could use a content check.

djarecka commented 1 month ago

Hi, I was using this PT to create a google sheet and create a linkml model.

A couple things that I noticed:

I think sub domain name could be simply "Project", "Person", etc. instead of "Project Metadata", "Person Metadata", etc.
in the file with DataCollection the sub domain name is "Project Metadata"
for all metadata there are two columns: definition and short description. I'm not completely sure what is teh meaning of short description. Often this is almost the same as definition, sometimes is enough different that I wouldn't guess that they describe the same field
assuming that if data type has "[]", that means that it is multivalued
it is not completely sure how should I treat "Person.name | Organization.name" in the data type field, name is just a string
line 8, 9, 10 in DataCollection file have are about accessControl, two of them have "enum" in data type, but without specifying the name of enum. I have only one ValueSet, that is called "accessControl" with values only, no description, no extra information, so not sure how would you like to map this

pbishwakarma commented 1 month ago

Hi Dorota,

Thanks for the comments, this had totally fell off my radar. I will update this doc based off your comments and some review.

Yes I agree subdomain can just be the entities
I will correct this
From the donor-to-alignment schema looks like short description is meant to provide some more information than the outright definition, maybe some examples. I will update
Yes [] should be multivalued
That is a polymorphic type - so in the case of a contact for a project, that can be either a Person or an Organization. The value that serves as the join key is the name of the Person or Organization entity.
accessControl right now is a simple enum (open/controlled access), what we'd like to do is evolve acccessControl to be composed of three things (some code to categorize the accessControl, a label for that code, and a description to allow people to specify reasons for why a subset if open/controlled). We haven't made this update yet, so if this model should be representative of current state, I think this should be a single enum field.

djarecka commented 1 month ago

@pbishwakarma - btw. I was also trying to identify the elements that I could find in the Data Catalog (or nemo website that DC points to) and in the google doc that I used to create linkml model from I created tab Checks: Slots to make some notes (see column k and l): https://docs.google.com/spreadsheets/d/1dCFxzpLBReauJYIinCs_xZlTckpJWFx5nxSTFAe-vvE/edit?usp=sharing

Happy to meet next week if you're planning to work on it and want to discuss. I was also told to look at the Dandi model and see what things could be also taken from there if useful.

pbishwakarma commented 1 month ago

@pbishwakarma - btw. I was also trying to identify the elements that I could find in the Data Catalog (or nemo website that DC points to) and in the google doc that I used to create linkml model from I created tab Checks: Slots to make some notes (see column k and l): https://docs.google.com/spreadsheets/d/1dCFxzpLBReauJYIinCs_xZlTckpJWFx5nxSTFAe-vvE/edit?usp=sharing

Got it, there are definitely some parts of the data model that are not being used currently (either they contain no data or just aren't being displayed in the UI) - I will update this PR before the working meeting on Wednesday if you want to discuss specifics about those nuances then.

Happy to meet next week if you're planning to work on it and want to discuss. I was also told to look at the Dandi model and see what things could be also taken from there if useful.

I think we took a stab at a similar exercise in the BICCN repo a while back, at least for mapping fields from our model to the archives' models, might be useful: https://github.com/BICCN/BCDC-Metadata/blob/83b00c19d21a7ef73196936a7b8e66b8637e59ee/design/schema/mappings.csv

djarecka commented 1 month ago

Got it, there are definitely some parts of the data model that are not being used currently (either they contain no data or just aren't being displayed in the UI) - I will update this PR before the working meeting on Wednesday if you want to discuss specifics about those nuances then.

Great, thank you!

I think we took a stab at a similar exercise in the BICCN repo a while back, at least for mapping fields from our model to the archives' models, might be useful: https://github.com/BICCN/BCDC-Metadata/blob/83b00c19d21a7ef73196936a7b8e66b8637e59ee/design/schema/mappings.csv

yes, I think I even created the column for Dandi :) it was just before I knew anything about BICCN or BICAN

djarecka commented 1 month ago

btw., perhaps you could update in google doc in the Slots tab, since I already fixed some of the issues: https://docs.google.com/spreadsheets/d/1dCFxzpLBReauJYIinCs_xZlTckpJWFx5nxSTFAe-vvE/edit?usp=sharing

Also, for the model, I would need more about the Classes

brain-bican / metadata-schemas

Add project, data collection metadata and associated CV tables #22