#=======================================================================
#Example code for how to make a violin graph in ggplot
#Example data pulled from paleobiology final exam
#Haleh Mawson, halehmawson@gmail.com
#Dec 2020
#=======================================================================
path<-"C:\\Users\\haleh\\Documents\\Biol 438 Paleobiology\\Take Home Final\\MA.vio.csv"
MA.vio <- read.csv(path) #Data is leaf mass per area and time period
#Goal is to show difference in means and dist. of LMA between two periods 
#(K & Pg)

#install ggplot2 if you don't have it
library(ggplot2) #load ggplot

#Build the base plot
vp <- ggplot(data = MA.vio, aes(x = Time, y = MA)) + 
  geom_violin(trim = F, fill = "lightblue") 
#ggplot() supplies the data. Make sure it is formatted correctly
#Should be in two or more adjacent columns, with no named rows
#geom_violin() creates the graph
#trim = F means that the plot will have gradiated edges, not sudden cut-offs
#fill = color

vp #check plot to make sure it worked

#Add bells and whistles
vp + stat_summary(fun.y=mean, geom="point", shape=20, size=4, col = "blue") +
  theme(panel.background = element_blank()) +
  ggtitle("Leaf mass per area by period") + ylab("Leaf mass per area (g/m2)")
#stat_summary() adds mean (or median, or whatever)
#theme() gets rid of gray background
#ggtitle() sets main title
#ylab() relabels y axis

#For adding quartiles, include a mini boxplot in the middle
vp + geom_boxplot(width=0.1, fill = "lightblue") + 
  theme(panel.background = element_blank()) +
  ggtitle("Leaf mass per area by period") + ylab("Leaf mass per area (g/m2)")
#These boxplots are awful because the data is awful

#Or add mean and sd
#install Hmisc package and it reqs
library(Hmisc)
vp + stat_summary(fun.data=mean_sdl, geom="pointrange", color="darkblue") + 
  theme(panel.background = element_blank()) +
  ggtitle("Leaf mass per area by period") + ylab("Leaf mass per area (g/m2)")
#Now you can see again how ugly the data is

#Change colors by group
vp <- ggplot(data = MA.vio, aes(x = Time, y = MA, fill = Time)) + 
  geom_violin(trim = F) +
  theme(panel.background = element_blank()) +
  ggtitle("Leaf mass per area by period") + ylab("Leaf mass per area (g/m2)")
vp
#Now the periods are different colors
#The default title is left aligned