Staging gEAR-Request failed with status code 500

[beamilon]

@adkinsrs 3 datasets (ScRNA-seq - P8, P12, P20 mouse cochlea (Michalski 2023; SnRNA-seq - P8 mouse cochleae (Michalski 2023; Mouse, scRNAseq,postnatal mouse utricle,raw matrix (Jan)) ), from the curated profiles do not show the results and the following message appears: Request failed with status code 500.

[beamilon]

@adkinsrs @JPReceveur There are more datasets with the error:

• "in vitro, chick, RNA-seq, utricle, response to Streptomycin (Lovett)" in Chick
• "Chicken Basilar Papilla Baseline, P7, violin plot, scRNA-seq (Janesick), geneID"; "HC Tonotopy Dynamic Expression Plot, P7, HC States, scRNA-seq (Janesick), geneID"; "HC Tonotopy Dynamic Expression Plot, P7, Tall HC Trajectory, scRNA-seq (Janesick),updated, geneID"; "HC Tonotopy Dynamic Expression Plot, P7, short HC,scRNA-seq (Janesick),geneID" in scRNA-seq - Chicken basilar papilla (Heller 2021)

[adkinsrs]

@beamilon I don't think staging.umgear.org is up anymore. Are you referring to the main umgear.org now? Also I have been fixing various bugs but have not updated the codebase on the server in a while. I just updated the codebase, though I think none of the updates would address your issues, but it would be nice to have a confirmation.

[beamilon]

Yes, I am using the main gEAR. Some datasets also come up with a "Could not retrieve analysis."

[adkinsrs]

The "could not retrieve analysis" means that the default curation (user or owner) was saved using data from a saved analysis. However, in cases where it was a "user_saved" analysis, if you were not the user who saved that analysis, then the analysis cannot be retrieved because it was another user's (and that would violate privacy)

[beamilon]

These datasets were fine in gEAR v1. Is it a new code?

[adkinsrs]

I think gEAR v1 never honored if the default display originated from the original dataset or an analysis, instead always using the original dataset. And a potential issue with going back to the original dataset may be that someone using something specific to the analysis like clustering groups in their curation that is not found in the analysis.

[adkinsrs]

I'll take a further look in any case. This issue with analyses has already been discussed in other tickets.

[adkinsrs]

@adkinsrs @JPReceveur There are more datasets with the error:

• "in vitro, chick, RNA-seq, utricle, response to Streptomycin (Lovett)" in Chick
• "Chicken Basilar Papilla Baseline, P7, violin plot, scRNA-seq (Janesick), geneID"; "HC Tonotopy Dynamic Expression Plot, P7, HC States, scRNA-seq (Janesick), geneID"; "HC Tonotopy Dynamic Expression Plot, P7, Tall HC Trajectory, scRNA-seq (Janesick),updated, geneID"; "HC Tonotopy Dynamic Expression Plot, P7, short HC,scRNA-seq (Janesick),geneID" in scRNA-seq - Chicken basilar papilla (Heller 2021)

Fixed the errors with this.

First one, there was a bug where we weren't importing a package Second set, there were cases where the gene (I checked POU4F3) would map to multiple identifiers in the dataset (Entrez IDs in this case), and this broke some code I wrote to fix #878

[adkinsrs]

I can confirm with the Mouse, scRNAseq,postnatal mouse utricle,raw matrix (Jan) dataset that both saved displays use a "user_saved" analysis, which means that if it does not exist under the user viewing the page (me in this example), then it cannot be retrieved, as we cannot walk through other users' directories for potentially private analyses. I noticed that for this dataset the gEAR curator account created the analysis... are you logged into the curator account when attempting to view or in your own personal account?

[beamilon]

I was on my account when I checked the dataset. But I just logged into the curator and I get the same error message.

[adkinsrs]

I was on my account when I checked the dataset. But I just logged into the curator and I get the same error message.

Seems the error was actually tied to processing the h5ad if you are on the right user. I fixed that error, logged into the curator account, and was able to view the tSNE plot

[beamilon]

I confirm that it is working with the curator now. However, I thought that what was processed by the curator to help the author of a dataset would be made public and available, as long as the dataset is public of course.

[adkinsrs]

I confirm that it is working with the curator now. However, I thought that what was processed by the curator to help the author of a dataset would be made public and available, as long as the dataset is public of course.

Unfortunately, no. The analysis has its own private (user saved/user unsaved) and public (public/primary) settings independent of the dataset.

[adkinsrs]

We can discuss it in a meeting to let the curator-owned analyses always be public analysis, but it would require some work by @jorvis and/or myself. Personally, I would rather avoid having "exceptions to the rule" and just have analyses set to the right type individually