nfdi4plants / ARC-specification

12 stars 12 forks source link

Add `Image Directory` Node type for Input/Output #82

Closed kMutagene closed 1 month ago

kMutagene commented 7 months ago

I have seen there is an Image File Material Node

https://github.com/nfdi4plants/ARC-specification/blob/427935a6fe27ba611891e79027da4c86b82e1704/ISA-XLSX.md?plain=1#L626

maybe we can add Image Directory (or something along those lines) to allow for a single node that contains image series, e.g. for omero @HLWeil ?

HLWeil commented 7 months ago

Following an offline discussion about reducing (unneeded) specificity of the IO entity types:

How about we switch from

Raw Data File
Derived Data File
Image File

to

Data File
Data Directory

The reasoning would be the following:

  1. Currently, there is no way to annotate directories, which we will definitely need.
  2. Distinction between Raw and Derived is kind of arbitrary (and can possibly be derived from the process graph)
  3. Why is there Image File, but not Binary File or Text File or Video File or whatever. Questions regarding this level of specifity could be endless.
HLWeil commented 7 months ago

@Freymaurer @Brilator @muehlhaus

kMutagene commented 7 months ago

this is the way

Freymaurer commented 7 months ago

lgtm

kMutagene commented 7 months ago

Maybe even go a step further and default to Material Name in the data model when there is only Output without brackets?

Brilator commented 7 months ago

directories, which we will definitely need

why?

Strongly agree with the other two points.

kMutagene commented 7 months ago

directories, which we will definitely need

why?

While working through the imaging use cases, the need for treating a directory that contains a set of files as a single entity (meaning one output node) became obvious.