# import library for effect size test: (cohens-d)
library("effectsize")

#filename = "/data/HyphaeVsYeast/quantification/fungus-fraction_per-sample.csv"
filename = "/data/HyphaeVsYeast/quantification/fungus-fraction_per-sample_withCEA.csv"


data = read.csv(filename, dec=".") #",")

#data$FungusFraction <- as.numeric(data$FungusFraction)    
#data = data[data$Group=='24h' | data$Group=='72h',]

summary(data)

# 24h vs. 72h - Fungus - Time Point
boxplot(FungusFraction~Group, data=data)
wilcox.test(FungusFraction~Group, data=data, paired=FALSE)
cohens_d(FungusFraction~Group, data=data, paired=FALSE)

# 24h vs. 72h - Fungus - Genotyp
boxplot(FungusFraction~Genotyp, data=data)
wilcox.test(FungusFraction~Genotyp, data=data, paired=FALSE)
cohens_d(FungusFraction~Genotyp, data=data, paired=FALSE)

# filter the data for time point
data24 = data[data$Group=='24h',]
data72 = data[data$Group=='72h',]

# 24h - Fungus
boxplot(FungusFraction~Genotyp, data=data24)
wilcox.test(FungusFraction~Genotyp, data=data24, paired=FALSE)
cohens_d(FungusFraction~Genotyp, data=data24, paired=FALSE)

# 72h - Fungus
boxplot(FungusFraction~Genotyp, data=data72)
wilcox.test(FungusFraction~Genotyp, data=data72, paired=FALSE)
cohens_d(FungusFraction~Genotyp, data=data72, paired=FALSE)

########### CC6-AK - comparison ###########

# 24h - Fungus - CC6-AK vs. CEA
data24h_CC6AK_CEA = data[data$Group!='72h' & data$Genotyp != 'WT',]

boxplot(FungusFraction~Genotyp, data=data24h_CC6AK_CEA)
wilcox.test(FungusFraction~Genotyp, data=data24h_CC6AK_CEA, paired=FALSE)
cohens_d(FungusFraction~Genotyp, data=data24h_CC6AK_CEA, paired=FALSE)

# 24h - Fungus - CC6-AK vs. WT
data24h_CC6AK_WT = data[data$Group!='72h' & data$Genotyp != 'CEA',]

boxplot(FungusFraction~Genotyp, data=data24h_CC6AK_WT)
wilcox.test(FungusFraction~Genotyp, data=data24h_CC6AK_WT, paired=FALSE)
cohens_d(FungusFraction~Genotyp, data=data24h_CC6AK_WT, paired=FALSE)




# scatter plot
plot(data$FungusFraction, y=data$Group)




######## all data points grouped by Genotyp ########

# Fungus fraction of ALL valid data by Genotyp
boxplot(Fungus.fraction~Genotyp, data=dataALL)
wilcox.test(Fungus.fraction~Genotyp, data=dataALL, paired=FALSE)
cohens_d(Fungus.fraction~Genotyp, data=dataALL, paired=FALSE)

# Fungus fraction of ALL valid data by time
boxplot(FungusFraction~Group, data=dataALL)
wilcox.test(FungusFraction~Group, data=dataALL, paired=FALSE)
cohens_d(Fungus.fraction~Group, data=dataALL, paired=FALSE)

# Yeast fraction of ALL valid data grouped by Genotyp
boxplot(Yeast.fraction~Genotyp, data=dataALL)
wilcox.test(Yeast.fraction~Genotyp, data=dataALL, paired=FALSE)
cohens_d(Yeast.fraction~Genotyp, data=dataALL, paired=FALSE)

# Yeast fraction of ALL valid data grouped by time
boxplot(Yeast.fraction~Group, data=dataALL)
wilcox.test(Yeast.fraction~Group, data=dataALL, paired=FALSE)
cohens_d(Yeast.fraction~Group, data=dataALL, paired=FALSE)

######## T-tests ########

# T-test für 24h (nur für Normalverteilungen)
t.test(Fungus.fraction~Genotyp, data=data24, paired=FALSE)
t.test(Yeast.Fraction~Genotyp, data=data24, paired=FALSE)
t.test(Hypqe.Fraction~Genotyp, data=data24, paired=FALSE)

# T-test für 72h (nur für Normalverteilungen)
t.test(Fungus.fraction~Genotyp, data=data72, paired=FALSE)
t.test(Yeast.Fraction~Genotyp, data=data72, paired=FALSE)
t.test(Hypqe.Fraction~Genotyp, data=data72, paired=FALSE)