Welcome to Reproducible Research Network

A network of students and faculty encouraging reproducible research and data analysis practices

https://ucd-reproducible-research-clinic.github.io/

https://nickgeorge.net Presentation

https://ucd-reproducible-research-clinic.github.io/can-graphpad-do-this/p-value-plots.html

Can Graphpad do this?

Plots and code taken from:

• https://www.r-bloggers.com/add-p-values-and-significance-levels-to-ggplots/

GraphPad

• easy(ish) to use
• decent plots
• walks you through basic stats
• ???

plot

• we have time series or dose-response data and we want to plot it

statistics

• which ones are significant?

"Write the asterisks manually"

problem

$150-300/year license to copy paste asterisks?

How is this not figured out by now?

What about this?

• What if you could generate a chart that looks like this with a few lines of code?

ggpubr

• R programming language

Add on libraries:

• ggplot2
• ggpubr

Free as in beer, Free as in speech

How?

library(ggplot2)
library(ggpubr)
 str(ToothGrowth)

'data.frame':	60 obs. of  3 variables:
 $ len : num  4.2 11.5 7.3 5.8 6.4 10 11.2 11.2 5.2 7 ...
 $ supp: Factor w/ 2 levels "OJ","VC": 2 2 2 2 2 2 2 2 2 2 ...
 $ dose: num  0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 ...

• This dataframe (spreadsheet) has two treatments (Orange Juice, Vitamin C)
• contains a dose and measured length of teeth

Plot

This is as easy as:

library(ggplot2)
library(ggpubr)

data("ToothGrowth")

ggline(ToothGrowth, x = "dose", y = "len", add = "mean_se",
          color = "supp")+
  stat_compare_means(aes(group = supp), label = "p.signif", 
                     label.y = c(16, 25, 29))

For the multiple comparisons,

# specify comparisons
 my_comps <- list(c("0.5", "1"), c("1", "2"), c("0.5", "2"))

 ggboxplot(ToothGrowth, x = "dose", y = "len",
           color = "dose")+ 
   stat_compare_means(comparisons = my_comps,
                      label.y = c(29, 35, 40))+
   stat_compare_means(label.y = 45)

How can I do this with my data?

come to RRN meetings!

How is this reproducible?

You go from your RAW data to this plot

• all steps are recorded
• all steps are automated
• no copy paste mistakes!
• prettier plots!

Links

Plots and code taken from:

• https://www.r-bloggers.com/add-p-values-and-significance-levels-to-ggplots/

Spread the word and let us help you!

• https://ucd-reproducible-research-clinic.github.io/