Consider alternative method to display distributions in Data Explore section

[enemeth19]

i.e., instead of pie charts

[rmoffitt]

It is nice that the pie charts show the frequency of some of these mutations- it doubles as a visualization for the user. You're right though, that there's not much of a benefit to the pie chart when the goal is to select subgroups.

[spencerkrichevsky]

i.e., instead of pie charts

what do you think about stacked barplots? the x-axis could be gene names, the y-axis could be counts, and the color of each stack could be gene signature or clinical metadata

fake data example:

[enemeth19]

what do you think about stacked barplots?

This seems like an improvement to me. Thoughts? @19aanand @rmoffitt

[adit-anand]

I'm for this since it makes the raw counts easier to discern! To clarify, would the idea be to have a single, independently-rendered stacked bar for each clinical feature or gene?

[enemeth19]

would the idea be to have a single, independently-rendered stacked bar for each clinical feature or gene?

Yup, that was the idea (for genes)

[adit-anand]

Update from today's WebGen meeting: For the version of the app we use in the manuscript that gets submitted, we will continue using pie charts for the Data Explore section. Future versions of the app may implement a new version of showing counts.

Should we close this issue and reopen it when we are focusing on adding new features?

[enemeth19]

Should we close this issue and reopen it when we are focusing on adding new features?

Sounds fine with me. We could do that or give it an "on hold" label

[enemeth19]

Thoughts on addressing this issue before we publish the app? The more I think about doing so, the more it makes sense to me to serve our mission of facilitating intuitive/easy-to-use data exploration and cohort selection