Analytics (new): <Diff Expression Feature>

[niknak33]

Overview

Tool: <Differential Expression>

• Feature: <Collection of all features within the new Differential Expression Tool>

Description: <New tool being built for CELLxGENE suits of products, this will be a new tab and therefore a new analytics request is being created> Expectations: <Potentially high usage for this tool, this is because we have had requests for this tool over the past few years.>

Learning questions

• Are users actually using the diff button?
• Are users often comparing healthy data to disease data?
• Are people coming from other product areas to Diff Exp?
• Are people going back to other product areas after performing Diff Exp?
• Is the first page of genes in the results panel the most salient or are users searching for other genes?

What to make sure to measure

• Measure users who come from Explorer
• Measure users who come from Gene Expression
• Measure users who make a change from the filters/selections made in Gene Expression
• Measure users who come via the DE tab
• Measure users who make a selection to any filter on DE
• Measure users who run/execute DE
• Measure users who download the results
• Measure users who Open Gene Expression from DE

New events to add

Differential expression DE_DOCUMENTATION_CLICKED

• visitor clicked the documentation button with the tool description at the top

DE_SELECT_CELL_TYPE, DE_SELECT_TISSUE, DE_SELECT_DISEASE, DE_SELECT_SEX, DE_SELECT_ETHNICITY, DE_SELECT_PUBLICATION : {<field based on category>: 'value selected (name)'}

• visitor made a selection for any of the categories for Cell group 1 or Cell group 2
• Example event: DE_SELECT_CELL_TYPE: {cell_type: 'alveolar macrophage'}

DE_CG_COPY_CLICKED: {category: 'category button clicked for (i.e. cell type, tissue, etc.)', values: 'string of IDs of values in Cell group 1 when copy was clicked' }

• visitor clicked to copy their selections for a category in Cell Group 1 to Cell Group 2
• Example event: DE_CG_COPY_CLICKED: {category: 'tissue', values: 'UBERON:xxxx,UBERON:xxx'}

DE_CG_DIFF_CLICKED: {category: 'category button clicked for (i.e. cell type, tissue, etc.)', values: 'string of IDs of values in Cell group 1 when diff was clicked' }

• visitor clicked to add the inverse of what's in Cell Group 1 to Cell Group 2 for a specific category
• Example event: DE_CG_DIFF_CLICKED: {category: 'tissue', values: 'UBERON:xxxx,UBERON:xxx'}

DE_FIND_GENES_CLICKED : {<field based on category>: 'CG1: <values selected (IDs) for cell group 1>, CG2: <values selected (IDs) for cell group 2>',...}

• visitor clicked to calculate the differential expression

• Example event: DE_FIND_GENES_CLICKED: {cell_type: 'CG1: CL:0000583; CG2: none', tissue: 'CG1: UBERON:0002048; CG2: UBERON:0002048', disease: 'CG1: PATO:0000461; CG2: MONDO:0100096, MONDO:0009061'}

  DE_DOWNLOAD_CLICKED : {<field based on category>: 'CG1: <values selected (IDs) for cell group 1>, CG2: <values selected (IDs) for cell group 2>',...}

• visitor clicked to download the results
• Example event: DE_FIND_GENES_CLICKED: {cell_type: 'CG1: CL:0000583; CG2: none', tissue: 'CG1: UBERON:0002048; CG2: UBERON:0002048', disease: 'CG1: PATO:0000461; CG2: MONDO:0100096, MONDO:0009061'}

DE_SEARCH_GENE : {gene: 'gene searched for'}

• visitor searched for a gene in the results panel

Cell Guide CG_OPEN_IN_DE_CLICKED: {cell_type: 'contains which cell type page the visitor is on', tissue: 'contains what tissue filter was selected}

• visitor clicked to open in differential expression one of the cell type pages

Explorer EXPLORER_OPEN_DE_CLICKED

• visitor clicked to open differential expression tool in the pop up bar

Gene expresison WMG_OPEN_IN_DE_CLICKED: {cell_type: 'contains which cell type sidebar the visitor is on', tissue: 'contains what tissue that cell type was selected for'}

• visitor clicked to open in differential expression from the cell type side bar

Checklist

• [ ] Add events to the custom goals section of both the staging and prod Plausible spaces
• [ ] If rdev environment is created for this new feature, once it has been deployed, ping @ainfeld to check analytics
• [ ] Once feature is deployed to Staging ping @ainfeld to check analytics again
• [ ] Verify events are showing up in the Staging space in Plausible correctly
• [ ] Once feature is deployed to Prod ping @ainfeld
• [ ] Confirm Plausible event names in Prod look the same as Staging and are showing up in the Prod space in Plausible correctly

People to cc

PM: DS: @ainfeld

Privacy policy: https://cellxgene.cziscience.com/privacy

[ainfeld]

Updates to instrumentation DE_SELECT_CELL_TYPE, DE_SELECT_TISSUE, DE_SELECT_DISEASE, DE_SELECT_SEX, DE_SELECT_ETHNICITY, DE_SELECT_PUBLICATION

• currently: the event tracks all selected values in the current state of the diff exp landscape
  • although it's good to track the combination of values selected I think this is better to be done once the user actually runs the differential expression so we know what the end state of each category was
• update needed: just the current value being selected needs to be tracked
  • e.g. (1) select adipose tissue - DE_SELECT_TISSUE : {tissue: 'adipose tissue'} then (2) select adrenal gland with adipose tissue still selected - DE_SELECT_TISSUE : {tissue: 'adrenal gland'}
  • should just be the singular names i.e. cellType, publication, disease, etc.
  • this will help us more easily identify the most commonly selected values for each category

DE_FIND_GENES_CLICKED and DE_DOWNLOAD_CLICKED

• currently: includes all values in the current state of the diff exp landscape
• updates needed
  • [x] just include the non-greyed out values for each category
    • cells of those values aren't really being included in the calculation so we shouldn't be including them in our tracking either
  • [x] I wrote this incorrectly in the Zenhub issue, please include a ";" character between the CG1 and CG2 sections for each category
    • e.g. DE_FIND_GENES_CLICKED : {cellTypes: "CG1: CL:0000136,CL:0000236; CG2: CL:0000115"}
    • easier to split in post processing
  • [x] sort the string of values for each category alphabetically
    • want to be able to count number of visitors with each unique combination of values no matter what order they were added in

DE_SEARCH_GENE

• currently: fires every time a new letter is typed
• update needed: no need to track what is being typed in, just fire whenever anything is being typed in
  • it would be difficult to tell what exactly they were looking for anyways

New instrumentation needed given UI updates DE_OPEN_IN_GE_CLICKED

• visitor clicked to launch in gene expression for either cell group 1 or cell group 2

[ainfeld]

In addition to the changes needed in the comment above:

DE_CG_COPY_CLICKED and DE_CG_DIFF_CLICKED events can now be removed as the buttons were removed

DE_FIND_GENES_CLICKED

• if an overlap behavior is selected, update to add a field to the payload overlap and include as the value what the overlap selection was

Please let me know when all updates have been made and I can check it all one more time before launch. Thanks!!