# Please do not hesitate to contact either Dan (Dan.Nussey@ed.ac.uk) or Rachael (Rachael.Wilbourn@ed.ac.uk) if you encounter a problem 
# with this script. 

#=========================================================================================================================================

# The cox proportional hazards model computes the multiplicative effect of the included covariates on the estimated baseline hazard/
# mortality rate. The models below estimate the multiplicative effect of telomere length on baseline mortality rate. Telomere length will 
# first be transformed using a standard normal transformation, defining a standard deviation difference as the unit of analysis, which is 
# needed in order to standardise TL variation among studies. Before you can run the models you will first need to put together a data file 
# in excel that includes as many of the following variables as possible:

# Individual_ID
# start_time (this is the moment/year of blood sampling)
# end_time (this is the moment/year the study stopped the following of individuals)
# survival_status (status at the end_time of the survey: 0=survived; 1=disappeared)
# telomere_length (for qPCR the t/s ratio values are preferred)
# sex (0=female; 1=male)
# age_at_blood_sampling

#Please make sure that you have used the same variable headers as listed above and then save the excel file using the file format *.txt

#=========================================================================================================================================

#The code below will load the data. You will need to specify the data directory locating your file.

####IMPORTANT####
# 1 Please match the decimal character you used in excel with the code below. Currently the decimal is the default "." If you used a "," 
# you will need to specify that in the code below.
# 2 Please write an NA in every empty cell in your excel file. This will make sure that R understands that these cells are empty. 



rm(list = ls(all = TRUE))

telodata<-read.table(file="XXXXX",header=TRUE,dec=".",na.strings=c("NA"),sep="")
str(telodata)

#This will load the required survival library
library(survival)

#Here telomere length will be transformed using the standard normal transformation
telodata$TL<-(telodata$telomere_length-mean(telodata$telomere_length))/sd(telodata$telomere_length)
                  
#statistical models computing the hazard ratio of standardized telomere length
## with mortality data
model1<-summary(coxph(Surv(start_time,end_time,survival_status) ~ TL, data=telodata))
model1



######## IF YOUR DATA CONTAINS LIFESPAN/ CENSORED DATA ########

# please include the following variables and run the following code and alternative model 1

# Individual_ID
# surv_time (the survival time of the individual)
# censor (censored data 0 = alive, 1 = dead)
# telomere_length (for qPCR the t/s ratio values are preferred)
# sex (0=female; 1=male)
# age_at_blood_sampling

rm(list = ls(all = TRUE))

telodata<-read.table(file="XXXXX",header=TRUE,dec=".",na.strings=c("NA"),sep="")
str(telodata)

#This will load the required survival library
library(survival)

#Here telomere length will be transformed using the standard normal transformation
telodata$TL<-(telodata$telomere_length-mean(telodata$telomere_length))/sd(telodata$telomere_length)

#statistical models computing the hazard ratio of standardized telomere length with mortality data
model1<-summary(coxph(Surv(surv_time,censor) ~ TL, data=telodata))
model1

#=========================================================================================================================================


#### SEX SPECIFIC MODELS ####

#subset original dataframe by sex
males <- subset(telodata, sex==1)
females <- subset(telodata, sex==0)

#statistical models computing the hazard ratio of standardized telomere length in males
model1b<-summary(coxph(Surv(start_time,end_time,survival_status) ~ TL, data=males))
model1b

#statistical models computing the hazard ratio of standardized telomere length in females
model1c<-summary(coxph(Surv(start_time,end_time,survival_status) ~ TL, data=females))
model1c


#=========================================================================================================================================