README.Rmd

title: "TPH2 KO statistics"
output: github_document
author: "Alex Meng, Joanes Grandjean"

knitr::opts_chunk$set(echo = TRUE, eval=FALSE,warning = FALSE,message=FALSE)

Setup environement

Download DABEST

# you just need to run these once. 

install.packages('devtools')
install.pacakges("reshape2")
install.pacakges("wesanderson")

devtools::install_github("ACCLAB/dabestr")   
devtools::install_github("karthik/wesanderson")

Load pacakges

library(tidyverse)
library(glue)
library(dabestr)
library(wesanderson)  # see https://github.com/karthik/wesanderson
library(reshape2)
pal <- wes_palette("Darjeeling1")

Figure 1: Anxiety level in THP2. if necessary, show rest in table

A. experimental diagram B. EPM time in open arm C. EPM total distance

Figure 2: Social behaviour.

A. experimetnal diagram B. Total no contact C. Total no contact over time (as line plot) D. Total mounting % E. Total aggressive %

Figure 3: SHow gene expression left, protein right, anat center? rest is in table.

A. Brain punches location (color-code gene expression / protein) B. IL/PL C. centroid amygdala D. ventral CA1 E. Dorsal raphe

demo dabest in R

#load the table. for other cases, it might help to keep naming consistent between tables, see example. Also, avoid spaces or weird characters in table names. 
df <- read_csv('assets/tables/testname_measure.csv', col_types = cols()) %>% melt() %>% dplyr::rename(group = variable) %>% drop_na()

df %>% group_by(group) %>% summarise(mean=round(mean(value),2), sd=round(sd(value),2))

# estimate the p-values using non-parametric test
p_tmp<- c(wilcox.test(df$value[df$group == 'Tph2+/+'], df$value[df$group == 'Tph2+/-'])$p.value, wilcox.test(df$value[df$group == 'Tph2+/+'], df$value[df$group == 'Tph2-/-'])$p.value)

# estimate hedges'g
dabest_hedges <- dabest(df, group, value, idx = c('Tph2+/+','Tph2+/-','Tph2-/-'), paired = FALSE)  %>% hedges_g() 

#make the plot and save to file
Fig1A<-plot(dabest_hedges, palette = pal,  rawplot.ylabel = "EPM open field [%]") 
ggsave('assets/figure/fig_Fig1A.svg', plot = Fig1A, device = 'svg',dpi = 300)

#outputs the file to show in the doc. 
Fig1A

#outputs table with added p-values for good measure
dabest_hedges$result %>% mutate(p = p_tmp)
 

what happens if dabest bugs?

#load the table. for other cases, it might help to keep naming consistent between tables, see example. Also, avoid spaces or weird characters in table names. 
df <- read_csv('assets/tables/testname_measure.csv', col_types = cols()) %>% melt() %>% rename(group = variable) %>% drop_na()



dabest_hedges <- dabest(df, group, value, idx = c('Tph2+/+','Tph2+/-'), paired = FALSE)  %>% hedges_g() 

#figure will require post-pocessing to make do. 
plot(dabest_hedges, palette = pal)



dabest_hedges <- dabest(df, group, value, idx = c('Tph2+/+','Tph2-/-'), paired = FALSE)  %>% hedges_g() 

#figure will require post-pocessing to make do. 
plot(dabest_hedges, palette = pal)


 

Finally, you can consider https://rpkgs.datanovia.com/ggpubr/reference/ggarrange.html to make composite figures in R. Happy to help you get started.