You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Ability to add Table(s) to Coordinate Systems #443
Thanks for working on multi table support! I've been using the recent multi_table branch and it's been working pretty well so far. One nice feature would be the ability to add a table to coordinate system with a transformation.
For my particular use case, the purpose of this would be to group all related components of a particular field of view together for easy iteration / accession.
Hi @srivarra, great to hear that you tried already the multi_table branch and that you found it useful 😊
Regarding allowing tables to have an associated coordinate system, this would clash with the case in which a table is annotating multiple elements, each belonging to a different coordinate system.
Anyway, I can see two approaches to handle your use cases:
if you filter the SpatialData object with SpatialData.filter_by_coordinate_system(), the new object will subset the table to match the geometries in the coordinate system. The function was not in the docs, I will update this with a PR I am currently working on.
we plan to support NGFF files with arbitrary hierarchies Support for nested hierarchies in NGFF stores #398; the user will then be able for instance to create zarr group "sample0", "sample1" and to group all the elements and annotations belonging to a sample. This would not change how the SpatialData object and elements operate in-memory, and the use of coordinate systems would also not change.
Regarding allowing tables to have an associated coordinate system, this would clash with the case in which a table is annotating multiple elements, each belonging to a different coordinate system.
Yup, makes perfect sense to me.
I think the first approach sounds good for now, but the arbitrary hierarchy NGFF sounds perfect!
The function was not in the docs, I will update this with a PR I am currently working on.
Actually it was already in the docs but a bit hidden, since it's a method of the SpatialData class. We can make it more discoverable, tracking this in #225.
Activity
LucaMarconato commentedon Jan 29, 2024
Hi @srivarra, great to hear that you tried already the
multi_tablebranch and that you found it useful 😊Regarding allowing tables to have an associated coordinate system, this would clash with the case in which a table is annotating multiple elements, each belonging to a different coordinate system.
Anyway, I can see two approaches to handle your use cases:
SpatialDataobject withSpatialData.filter_by_coordinate_system(), the new object will subset the table to match the geometries in the coordinate system. The function was not in the docs, I will update this with a PR I am currently working on.SpatialDataobject and elements operate in-memory, and the use of coordinate systems would also not change.Would any of these work for your case?
srivarra commentedon Jan 29, 2024
@LucaMarconato
Yup, makes perfect sense to me.
I think the first approach sounds good for now, but the arbitrary hierarchy NGFF sounds perfect!
Thank you!
LucaMarconato commentedon Jan 29, 2024
Happy to hear this and looking forward to enable nested hierarchies! 😊 Closing this issue now as the progress will be tracked in #398.
LucaMarconato commentedon Feb 1, 2024
Actually it was already in the docs but a bit hidden, since it's a method of the
SpatialDataclass. We can make it more discoverable, tracking this in #225.