Closed milen-sage closed 1 year ago
To clarify the above, I am only able to use --use_schema_label
with submit
. I am not able to use get
to generate a manifest with schema labels as columns headers, and if I manually create a CSV with schema labels as column headers, validate
will throw an error. So I am forced to use display names in R, which is not ideal.
Hey @lopierra,
Can you provide me with your model and a manifest with entries? I would like to validate and submit to test that everything is working as you are intending.
Thanks!
Hi @mialy-defelice - here is our model in YAML and JSON-LD.
I'm attaching a test manifest for the Study component.
An additional complication is that if I submit my own manifest with schemaLabels as headers, Synapse automatically capitalizes the first letter of each (so studyCode
becomes StudyCode
). Is there some way to suppress that, or do I need to open a ticket for Platform about this?
An additional complication is that if I submit my own manifest with schemaLabels as headers, Synapse automatically capitalizes the first letter of each (so
studyCode
becomesStudyCode
). Is there some way to suppress that, or do I need to open a ticket for Platform about this?
@lopierra Are you submitting through the python or R client (using a table upload) or are you using Schematic?
I am using schematic.
I am using schematic.
@lopierra Okay, this is intentional in schematic to conform to synapse standards. We convert all headers to PascalCase during the final submission step. Are you always using camelCase? You can open a ticket with us.
@milen-sage do they need to be Class Labels or can they be Property Labels?
Thanks - I will open a ticket. We chose camelCase as the standard for slot names in our LinkML model and PascalCase for the classes, so it would be good to have the option for submitting camelCase headers to Synapse.
Based on conversation with @lopierra this is not an urgent blocker and we'll resolve after refactoring schema parsers #857
This issue is based on a request by @lopierra that began as a slack conversation, which can be viewed here.
User Story: As a data curator I would like the choice to generate the manifest tables column headers with schema labels instead display names for programatic ease of use.
AC:
--use_schema_label
as the column header when generating a manifest. Calling this label will set it toTrue
, default isFalse
and will default to using the display name.--use_schema_label
used to generate the manifestConfirm the following: Use several models -- Example, HTAN, iAtlas