ebi-gene-expression-group / atlas-web-single-cell

Single Cell Expression Atlas web application
Apache License 2.0
5 stars 5 forks source link

Experiment's default plot option is not correctly displayed in SCXA #374

Closed ke4 closed 4 months ago

ke4 commented 9 months ago

We discussed the use case that should be implement for this with @upendrakumbham and @@pmb59. There are 2 separate condition that we have to fullfill:

Steps to reproduce:

  1. Go to Browse experiments tab in SCXA.
  2. Select Transcriptional programs of circuit assembly in the Drosophila visual system

Expected result: The middle Plot option is selected, as all the priority column in scxa_dimension_reduction table is 0.

Actual result: The 1st Plot option is selected.

You should also test it with E-ANND-3 in your local environment. Please load that experiment into your local environment , if you have not done it, yet. The instructions to follow: Add an AnnData experiment to our local SCXA

Please see below how to reproduce the bug with that experiment.


For historical reason I preserved the original ticket description and you can find it below.

Original title: E-ANND-3 experiment by default is not displaying plots in SCXA

Steps to reproduce:

  1. Go to Browse experiments tab in SCXA.
  2. Select Tabula Sapiens - Smart-seq2 dataset experiment

Expected result:

Experiment's data should show up with the plots diagram.

Actual result:

An error message appears on the screen.

Image

ke4 commented 9 months ago

Here is the log from Solr cloud when it tries to load E-ANND-3:

anndata_not_appearing.log

I need to investigate this more closely.

ke4 commented 8 months ago

It looks like that I had an issue previously with my local SCXA environment regarding the data in my local DB. I repopulated it and I managed to display E-ANND-3 although there is a problem with the default view. I attached the default view here:

Image

ke4 commented 8 months ago

If you go and select the only plot option, then the cell plots are correctly appear:

Image

ke4 commented 8 months ago

After some investigation it looks like the backend code is not correctly calculating the default Plot options for a specific plot type for a specific experiment.

The DB query that used by this method is not correct: https://github.com/ebi-gene-expression-group/atlas-web-single-cell/blob/7a6d042bc72e801452fd981627cd5d2bfa945c88/app/src/main/java/uk/ac/ebi/atlas/experimentpage/cellplot/CellPlotDao.java#L169

It does not take account that more than one values could be the same (in this case: 0).

lingyun1010 commented 7 months ago

I fixed the software bugs but in the meanwhile I found some errors in PostgreSQL database on dev. I took a screenshot on what we have in the scxa-dimension_reduction search results. The issue is we have different formats on some experiments with t-SNE and tsne, umap and UMAP and mixture of three as shown in the screenshot.

Image