gabrifc / raincloud-shiny

Shiny app for customizing Raincloud plots

Raincloud-shiny

Raincloud-shiny is a Shiny-powered web GUI interface to create RainCloudPlots. A demo is hosted [here](https://gabrifc.shinyapps.io/raincloudplots/).

The idea behind Raincloud plots was introduced by Micah Allen on his blog. My coworkers and I found it really interesting to display our data, so I made this shiny app to provide a smooth transition to R and ggplot.

More information about Raincloud plots can be found on the in the published article.

Citing RainCloudPlots

Please refer to the original RainCloudPlots source for citations and licenses.

Credits

The code of RainCloud-Shiny makes use / is inspired by the following open source projects:

• RainCloudPlots: The original code provided by the manuscript authors.
• RShiny: The shinythemes package provides some Bootstrap themes for use with Shiny.
• ShinyThemes: Shiny is an R package that makes it easy to build interactive web apps straight from R.
• shinycssloaders:Add loader animations (spinners) to Shiny Outputs in an automated fashion.
• shinyWidgets: An extension of the default widgets available in Shiny.
• Tidyverse: An opinionated collection of R packages designed for data science. This app makes use of the following packages to wrangle and display data:
  • Tidyr.
  • Glue.
  • Stringr.
  • ggplot2.
• geom_flat_violin.R: a ggplot2 extension to make one side flat violins.
• Cowplot: Create publication-ready figures.
• RColorBrewer: Provides Color Theme Palettes designed by Cynthia Brewer as described at Colorbrewer 2.0.
• ggsci: Scientific Journal and Sci-Fi Themed Color Palettes for ggplot2
• ggpubr: Makes it possible to automatically add p-values and significance levels.
• ggbeeswarm: Provides methods to create beeswarm-style plots using ggplot2.
• Hmisc: Miscellaneous utilities from Frank Harrell. Here used to calculate the 95% Confidence Intervals.

raincloud-shiny's Issues

Add normality tests

If we are testing parametric/non-parametric, we should check if data is normal first.

Move libraries needed to the modules

Calls to libraries should be in the module to make them completely independent.

Edit boxplot values

Boxplots are defaulting to median, 25 & 75% intervals, and 95% CI.
We should be able to change it to the mean and such.

Style the R code

Maybe use styler (https://github.com/r-lib/styler)

Save input options between sessions

It is not fun to repeat the same options every time we restart the app.

Modularize data tidying

This would allow to ask which columns to use prior to plotting to reduce burden on the server, and also to make grouped analysis with stacked violin and boxplots.

Sensible defaults for placeholder values

Change the hardcoded values such as X and Y position for statistics to make them relative to the dataset.
Something similar with the ggtitle.

Allow custom colors for specific geoms

We should be able to edit the colours of the plot in addition to the palette.

Add multitest correction

Pariwise comparisons are not being corercted for multiple comparisons.
Add the necessary multitest corrections:

• Tukey for All vs All
• Dunnett's for All vs Control
• Sidak for specific multiple comparisons.

Allow downloading the whole script

For reproducibility, customization and/or journal publication.

Fix Modular Plot Download

When using modules and passing the plot as a reactive variable, the plots are not updated after the first download.
This is not happening when the downloadHandlers are hardcoded, so I am guessing that this is a problem with reactivity, modules and ids but I am not knowledgeable enough with that to understand why the rendered plot is uploaded but not the downloaded one.

Palette picker

Hi,
Nice project!
Last release of shinyWidgets on CRAN use new version of bootstrap-select and it implies some differences with the old behaviour, to make the palette picker works again, you can add somewhere in your UI this line :

tags$style(".bootstrap-select .dropdown-menu li a span.text {width: 100%;}")

(no better solution for the moment)

And maybe remove padding in the content :

"<div style='width:100%%;border-radius:4px;background:%s;color:%s'>%s</div>"
# instead of 
"<div style='width:100%%;padding:5px;border-radius:4px;background:%s;color:%s'>%s</div>"

Cheers,

Victor