Cell Browser testing

Ahead of a Cell Browser software release, the UI should be thoroughly tested to ensure that new features and bug fixes haven't unintentionally impacted old functions.

UI testing

In general, for every function that has a keyboard shortcut or UI button, check that the corresponding drop-down menu button/link works.

Try one dataset of every type

• at the minimum, try a RNA-seq dataset, e.g. the good old cortex-dev https://cells-test.gi.ucsc.edu/?ds=cortex-dev, try to zoom, color by meta, color my gene.
• try one ATAC dataset, e.g. https://cells-test.gi.ucsc.edu/?ds=retina+hrca+atac+ac, try color by peak and search for a gene
• try one spatial dataset, e.g. https://cells-test.gi.ucsc.edu/?ds=brain-spatial-omics+p22-h3k27me3, check that image in the background loads

Basic UI Functions

• Zoom in and out, then jump back to the default 100% zoom. Do so using both the UI buttons and keyboard shortcuts.
• Check that you can change the layout using the "Layout" drop-down
• For a collection, check that you can change between datasets in the collection using the "Collection" drop-down.
• Check the pan, square-select, and select-to-zoom functions
• Check "Info" and "Open" buttons work.

Selecting cells

• Select cells using the variety of methods available:
  • Click a metadata value in the legend (e.g. clicking "glioma" to select all glioma samples).
  • Use box select
  • Use "Find cells" under the "Edit" drop-down:
    • Change each of the drop-down menus
    • Add multiple filter criteria
    • Confirm that filter criteria are properly applied. A good way to do this is using the histogram percentages that show up after making selecting some cells (e.g. in the treehouse/compendium-v10-polyA dataset, there are 755 glioma samples, ~40% of which come from female patients. If you apply filters to select only those that are both glioma and female, then only 755 * 0.4 = ~302 samples should be selected.)
• After selecting cells, hover a metadata field and confirm that the histogram shows up.

Legend and Coloring

• Clicking a "Cell Annotation" field should color cells based on this field. Ensure that the legend is updated with the values from the field you clicked on.
• Sort the legend by both name and frequency.
• Change the colors of the legend using the "Colors" drop-down.
• Change the color of a single value to a custom color by clicking on the color box next to the value in the legend.
• Use the drop-down at the top of the "Cell Annotations" tab to change annotation to color by.
• Color cells based on the expression of a gene using the search box.
• Ensure that any dataset/quick genes associated with a dataset show-up under "Dataset Genes".
• After you select a gene to color on, ensure it shows up in the "Recent genes" section (and that clicking on that same gene again doesn't cause it to show up in that list twice).

Marker genes

• Hover over a cluster label and display the top cluster marker genes.
• Click on a cluster label to view the marker genes for that cluster. Note that not all datasets have marker genes, so make sure that the dataset has marker genes associated with it before checking this.
• Check linkouts to BrainSpLMD, Eurexp, OMIM, HPO, etc. Note that not every set of marker genes has these.
• Click on a marker gene name and ensure it then colors cells based on the expression of that gene

Advanced functions

• Test all the functions under the "Edit", paying special attention to "Name selection" and "Find cells".
  • Check that you can add custom annotations using "Name selection" and that you can remove them using Tools>Remove custom annotations
• Enable the heatmap feature and split-screen features separately and in combination. Try turning them on/off in different orders/combinations (e.g. try turning on the heatmap first and then the split-screen, and vice versa).
• Hide cluster labels
• Check the links to Xena and the UCSC Genome Browser (link is also present on the marker genes page)

Command-line util testing

Since we use the various cb* command-line utilities throughout the wrangling process it's not super necessary to test them separately. If you reported a bug with a particular tool and an engineer has fixed that bug, then you should test the fix to ensure that it is resolved and no new issues have been introduced.