library(MASS)

# the complete dataset for trypanosome prevalence is included in the file "Isaac_et_al_additional_file_1_complete_trypanosome_prevalence_dataset.txt"
# the following procedure assumes that the file "Isaac_et_al_additional_file_1_complete_trypanosome_prevalence_dataset.txt" is located on the C drive (file location "C:/Isaac_et_al_additional_file_1_complete_trypanosome_prevalence_dataset.txt" as indicated below)

#load complete dataset:
Dataset <- read.table("C:/Isaac_et_al_additional_file_1_complete_trypanosome_prevalence_dataset.txt", header=TRUE, sep="\t", na.strings="NA", dec=".", strip.white=TRUE)

Dataset <- within(Dataset, {Fly_sex <- as.factor(Fly_sex)})									#convert numberic variable (Fly_sex) into factor
Dataset <- within(Dataset, {Collection_site <- as.factor(Collection_site)})							#convert numberic variable (Collection_site) into factor
Dataset <- within(Dataset, {tsetse_species <- as.factor(tsetse_species)})							#convert numberic variable (tsetse_species) into factor

#logistic regressions on the complete data set:

GLM.1 <- glm(All_trypanosomes ~ Collection_site+Fly_sex+(Collection_site*Fly_sex), family=binomial(logit), data=Dataset)	# logistic regression for all trypanosome tsetse_species
stepGLM.Alltryps <- stepAIC(GLM.1, direction="backward", steps=2)			 					# stepwise model selection based on AIC
stepGLM.Alltryps$anova														# display results of model selection
GLM.1.1 <- glm(All_trypanosomes ~ Fly_sex, family=binomial(logit), data=Dataset)						# selected Model: All_trypanosomes ~ Collection_site
summary(GLM.1.1)														# output results for selected model
exp(coef(GLM.1.1))														# compute the odd ratios for selected model

GLM.2 <- glm(T.congolense_subspp. ~ Collection_site+Fly_sex+(Collection_site*Fly_sex), family=binomial(logit), data=Dataset)	# logistic regression for T.congolense_subspp.
stepGLM.T.congolense_subspp. <- stepAIC(GLM.2, direction="backward", steps=2)							# stepwise model selection based on AIC
stepGLM.T.congolense_subspp.$anova												# display results of model selection
GLM.2.1 <- glm(T.congolense_subspp. ~ Collection_site, family=binomial(logit), data=Dataset)					# selected Model: T.congolense_subspp. ~ Collection_site
summary(GLM.2.1)														# output results for selected model
exp(coef(GLM.2.1))														# compute the odd ratios for selected model

GLM.3 <- glm(T.congolense_Savannah ~ Collection_site+Fly_sex+(Collection_site*Fly_sex), family=binomial(logit), data=Dataset)	# logistic regression for T.congolense_Savannah
stepGLM.T.congolense_Savannah <- stepAIC(GLM.3, direction="backward", steps=2)							# stepwise model selection based on AIC
stepGLM.T.congolense_Savannah$anova												# display results of model selection (selected Model: T.congolense_Savannah ~ Collection_site + Fly_sex + Fly_sex*Collection_site)
summary(GLM.3)															# output results for selected model
exp(coef(GLM.3))														# compute the odd ratios for selected model

GLM.4 <- glm(T.congolense_Forest ~ Collection_site+Fly_sex+(Collection_site*Fly_sex), family=binomial(logit), data=Dataset)	# logistic regression for T.congolense_Forest
stepGLM.T.congolense_Forest <- stepAIC(GLM.4, direction="backward", steps=2)							# stepwise model selection based on AIC	
stepGLM.T.congolense_Forest$anova												# display results of model selection
GLM.4.1 <- glm(T.congolense_Forest ~ Collection_site, family=binomial(logit), data=Dataset)					# selected Model: T.congolense_Forest ~ Collection_site
summary(GLM.4.1)														# output results for selected model
exp(coef(GLM.4.1))														# compute the odd ratios for selected model

GLM.5 <- glm(T.vivax ~ Collection_site+Fly_sex+(Collection_site*Fly_sex), family=binomial(logit), data=Dataset)			# logistic regression for T.vivax
stepGLM.T.vivax <- stepAIC(GLM.5, direction="backward", steps=2)								# stepwise model selection based on AIC
stepGLM.T.vivax$anova														# display results of model selection (selected Model: T.vivax ~ Collection_site + Fly_sex + Fly_sex*Collection_site)
summary(GLM.5)															# output results for selected model
exp(coef(GLM.5))														# compute the odd ratios for selected model

yankari <- subset(Dataset, subset=Collection_site==2)										# create a subset dataset from the complete dataset for the Collection_site Yankari only

yankari <- within(yankari, {Fly_sex <- as.factor(Fly_sex)})									#convert numberic variable (Fly_sex) into factor in the new created yankari dataset
yankari <- within(yankari, {tsetse_species <- as.factor(tsetse_species)})							#convert numberic variable (tsetse_species) into factor in the new created yankari dataset


GLM.6 <- glm(All_trypanosomes ~ tsetse_species+Fly_sex+(tsetse_species*Fly_sex), family=binomial(logit), data=yankari)		# logistic regression for all trypanosome tsetse_species in yankari
stepGLM.Alltryps_yankari <- stepAIC(GLM.6, direction="backward", steps=2)		 					# stepwise model selection based on AIC
stepGLM.Alltryps_yankari$anova													# display results of model selection 	
GLM.6.1 <- glm(All_trypanosomes ~ Fly_sex, family=binomial(logit), data=yankari)						# selected Model: All_trypanosomes ~ Collection_site
summary(GLM.6.1)														# output results for selected model
exp(coef(GLM.6.1))														# compute the odd ratios for selected model

GLM.7 <- glm(T.congolense_subspp. ~ tsetse_species+Fly_sex+(tsetse_species*Fly_sex), family=binomial(logit), data=yankari)	# logistic regression for T.congolense_subspp.
stepGLM.T.congolense_subspp._yankari <- stepAIC(GLM.7, direction="backward", steps=2)						# stepwise model selection based on AIC
stepGLM.T.congolense_subspp._yankari$anova											# display results of model selection 
GLM.7.1 <- glm(T.congolense_subspp. ~ Fly_sex, family=binomial(logit), data=yankari)						# selected Model: T.congolense_subspp. ~ 1
summary(GLM.7.1)														# output results for selected model
exp(coef(GLM.7.1))														# compute the odd ratios for selected model

GLM.8 <- glm(T.congolense_Savannah ~ tsetse_species+Fly_sex+(tsetse_species*Fly_sex), family=binomial(logit), data=yankari)	# logistic regression for T.congolense_Savannah
stepGLM.T.congolense_Savannah <- stepAIC(GLM.8, direction="backward", steps=2)							# stepwise model selection based on AIC
stepGLM.T.congolense_Savannah$anova												# display results of model selection
GLM.8.1 <- glm(T.congolense_Savannah ~ Fly_sex, family=binomial(logit), data=yankari)						# selected Model: T.congolense_Savannah ~ Fly_sex
summary(GLM.8.1)														# output results for selected model
exp(coef(GLM.8.1))														# compute the odd ratios for selected model

GLM.9 <- glm(T.congolense_Forest ~ tsetse_species+Fly_sex+(tsetse_species*Fly_sex), family=binomial(logit), data=yankari)	# logistic regression for T.congolense_Forest
stepGLM.T.congolense_Forest <- stepAIC(GLM.9, direction="backward", steps=2)							# stepwise model selection based on AIC
stepGLM.T.congolense_Forest$anova												# display results of model selection (selected Model: T.congolense_Forest ~ Collection_site)
summary(GLM.9)															# output results for selected model
exp(coef(GLM.9))														# compute the odd ratios for selected model

GLM.10 <- glm(T.vivax ~ tsetse_species+Fly_sex+(tsetse_species*Fly_sex), family=binomial(logit), data=yankari)			# logistic regression for T.vivax
stepGLM.T.vivax <- stepAIC(GLM.10, direction="backward", steps=2)								# stepwise model selection based on AIC
stepGLM.T.vivax$anova														# display results of model selection
GLM.10.1 <- glm(T.vivax ~ Fly_sex, family=binomial(logit), data=yankari)							# selected Model: T.vivax ~ Fly_sex
summary(GLM.10.1)														# output results for selected model
exp(coef(GLM.10.1))														# compute the odd ratios for selected model