Quote from email:

At some point, including IDs in the legend is not helpful and takes up space that would be better used to visualize the data (see attached images). Do you have thoughts on the maximum number of IDs to display in the legend? My thought is that if there are more than ten or so, we can just remove the legend altogether.

It would help to add a data validation step when uploading data. This would provide the user with some more useful error messages when uploading data (if there is something wrong with the data).

• Error on the PCA, UMAP, and TSNE tabs: "An error has occurred. Check your logs or contact the app author for clarification".
• Dataset replicating this error can be accessed (internal link) here.
• Error generated when running locally:

Warning: Error in mutate: Problem while computing Group = as.character(meta[unique(.data$SampleID)]).
✖ Group must be size 2113 or 1, not 0.
ℹ The error occurred in group 1: SampleID = 0.

There is a bug stemming from too many levels in the grouping variable. I think it is coming from lines 90, 139, 158, 280, and 337 of utils.R (see 7bdaf70). Having all these groups is going to make the legend kind of pointless.

Test data may be included with the Dockerfile, making it larger than it needs to be. Double check that it isn't being copied over.

Extend extract_values to pull kmeans identifiers when exporting data.

Related to issue #10

• Replace existing tests with server function testing
• Add general Snapshot-based tests

See also this article.

Getting an error when trying to download PCA loadings:

Error in table: all arguments must have the same length

It is coming from somewhere in lines 615 - 705 of app_server.R in 1542de2

We are getting this warning with devtools::test()

Warning (test-app.R:2:3): app loads
Navigation containers expect a collection of bslib::nav()/shiny::tabPanel()s and/or bslib::nav_menu()/shiny::navbarMenu()s. Consider using header or footer if you wish to place content above (or below) every panel's contents.
Backtrace:

1. testthat::expect_true(...)
   at test-app.R:2:2
2. JoesFlow::run_app()
   at test-app.R:3:4
3. JoesFlow::app_ui()
4. shiny::navbarPage(...)
5. bslib::page_navbar(...)
6. bslib::navs_bar(...)
7. bslib:::navbarPage_(...)
8. bslib:::buildTabset(...)
9. base::lapply(...)
10. bslib (local) FUN(X[[i]], ...)

Change input for number of clusters from slider to text (integer) input

The app sometimes fails with an unhelpful message, "Disconnected from the server", with the option to reconnect. I've checked for memory issues, and while it does seem to happen more often with larger data sets, the memory usage is still well below the limit every time.

My best guess is that it gets bogged down when performing background calculations and crashes. However, while it does crash after making parameter changes and clicking around the app, I haven't identified a reproducible scenario that will crash the app.

Tabular output for composition is also broken (related to #11 ).