search

DataBiosphere / azul

Metadata indexer and query service used for AnVIL, HCA, LungMAP, and CGP
Apache License 2.0
7 stars 2 forks source link

Question about metadata appearing in browser for dev test 10X project #2687

Closed kbergin closed 3 years ago

kbergin commented 3 years ago

Here, in analysis protocol column, it lists both the Optimus version and the 'optimus post processing' pipeline that was run I assume because both have analysis protocols. I'd ideally not want the post processing listed, but I can understand why it should be given it was run on the data. Just wanted to note the question and see others thoughts.

hannes-ucsc commented 3 years ago

I'd ideally not want the post processing listed.

Why?

My uneducated opinion is that we do want it listed. But in the end, the analysis team controls what is displayed, they could simply use the same protocol for the Optimus step and the post-processing. I'd be opposed to implementing custom logic to filter that value out in Azul or DB. That would be brittle, costly to maintain for my team and create an inconsistent UX when the underlying metadata doesn't match the Azul response or the DB display.

kbergin commented 3 years ago

Yea that’s fair. I think it’s fine for now. For this data type the post processing analysis does do some modification of the data (merges matrices and filters a bit) so it’s good for it to be listed. Ideally the post processing would only be for data ETL purposes and so then I don’t love seeing it listed.

Thanks!

On Fri, Jan 8, 2021 at 2:38 PM Hannes Schmidt notifications@github.com wrote:

I'd ideally not want the post processing listed.

Why?

My uneducated opinion is that we do want it listed. But in the end, the analysis team controls what is displayed, they could simply use the same protocol for the Optimus step and the post-processing. I'd be opposed to implementing custom logic to filter that value out in Azul or DB. That would be brittle, costly to maintain for my team and create an inconsistent UX when the underlying metadata doesn't match the Azul response or the DB display.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/DataBiosphere/azul/issues/2687#issuecomment-756955212, or unsubscribe https://github.com/notifications/unsubscribe-auth/ACIDCJDHW26PXCDOPVP377DSY5NKJANCNFSM4V2TC5YQ .

-- Kylee Degatano Senior Software Product Manager Data Generation Pipelines Data Sciences Platform, Broad Institute

kbergin commented 3 years ago

I think based on Hannes comment this can be closed - unless you do plan to do something in response to it!

hannes-ucsc commented 3 years ago

I think Azul behaves as designed in that its web service response accurately reflects the metadata. If only one protocol should be returned by Azul, the metadata would have to only mention one protocol, not two different ones.