Design a new dataset for testing automation

[seve]

We would benefit from adding/replacing our current testing dataset(pbmc3k) with a deeper and more diverse dataset. This could be enhanced with a joining of multiple existing datasets that would stress our typing system as well. Would require a refactoring of our hard-coded values and snapshots.

Ideally this dataset would eliminate the need to bring non-annotation related tests into the e2eAnnotations.test.js testing file.

Rough to-do list:

• Define dataset requirements/design
• Create dataset
• Fix broken tests due to hard-coded values/snapshots

Some cases that need to be covered by the updated dataset:

• Abstract column and value names(genes, categorical columns, categorical values)
  • Character count
  • Character types
  • Existing bugs this would catch: fix truncation edge cases #1637
• Large # of columns/values
  • This would test our overflow/truncation handling
  • Existing bugs this would catch: fix label selection on trunctated category lists #1646
• Cells with NaN/inf embeddings
  • Will test Ability to plot embeddings that NaN values #1559 when a fix is developed
• Large Size Dataset
  • Test loading times/ fetch times

Feel free to comment more ideas and I'll add to the list!

[signechambers1]

@seve @maniarathi is this still relevant? do you want to include as some part of https://app.zenhub.com/workspaces/single-cell-5e2a191dad828d52cc78b028/issues/chanzuckerberg/single-cell/44?

[maniarathi]

Good call! I added it to our smoketest epic!

[signechambers1]

@maniarathi the smoke test epic is closed, ok to close this ticket?

[maniarathi]

Beefing up our set of valid-but-unusual datasets is still a thing we need to work on to be honest and we had to trim our efforts in testing due to time constraints. I'd like to keep this open because the ideas listed above are good, but I'm not sure when to get to it. I'd mark it as a P3.