Predicting-remission-Escitalopram.R

#Predictive model from GENDEP data 
#Antidepressant drug-specific prediction of depression treatment outcomes from genetic and clinical variables.
#Raquel Iniesta raquel.iniesta@kcl.ac.uk

#Use the model to estimate the probability of remission (yes or not) for a patient of depression after 12 weeks of treatment with Escitalopram.
#List of baseline predictors that are necessary for the prediction: 

#CLINICAL variables:
#appetite defined as in R. Uher et al. Depression symptom dimensions as predictors of antidepressant treatment outcome: replicable evidence for interest-activity symptoms Psychol. Med., 42 (2012), pp. 967–980
#BDI_changes_sleep is the score for BDI item number
#HRSD_13 is the score for HRSD-17 item number 13
#interest_activity_dimension defined as in R. Uher et al. Depression symptom dimensions as predictors of antidepressant treatment outcome: replicable evidence for interest-activity symptoms Psychol. Med., 42 (2012), pp. 967–980
#HRSD_total is the total score for HRSD-17
#SCAN_fatigability is the the score for SCAN item fatigability

#GENETIC variables: SNP variants coded as 0,1,2 where 0:Homozygous for minor allele; 1:Heterozygous; 2:Homozygous for major allele
#rs1392611 
#rs10812099 
#rs1891943 
#rs151139256 
#rs11002001 
#rs62182022 
#rs28373080 
#rs7757702 
#rs76557116 
#rs9557363 
#rs2704022 


#Install the latest version of R from https://cran.r-project.org 

#Copy and paste instructions from STEP 1 and STEP 2 in the R console:

#######STEP 1. Data update.
#Please supply the values for a specific patient. You must replace 0 values by the current patient measurements.
#Then copy and paste in the R console.

appetite<-0 
BDI_changes_sleep<-0  
HRSD_13<-0  
interest_activity_dimension<-0  
HRSD_total<-0  
SCAN_fatigability<-0  
rs1392611<-0 
rs10812099<-0 
rs1891943<-0  
rs151139256<-0 
rs11002001<-0 
rs62182022<-0 
rs28373080<-0 
rs7757702<-0 
rs76557116<-0 
rs9557363<-0 
rs2704022<-0 


#######STEP 2. Computation of probability of remission 
#Next code lines allow to estimate the probability of remission for an indidivual patient. 
#Please do not modify any value in the code!! Copy and paste in the R console.
#The final computation for the probability of remission will be stored in the object "prob.remission".

#do not modify!!
model.coefficients<-c(-0.018595926,-0.036970804,-0.037494038,-0.003203991,-0.033419381,-0.035164398,-0.015327153,-0.029090814,-0.032467316,
                      -0.017888137,-0.037494134,0.032473920,-0.019367106,0.020377797,0.017466538,0.032473995,-0.021702248,0.039596562)

#do not modify!!
predictors_mean_ingendep<-c(0.8357628,1.6223776,1.5734266,0.6386217,24.1856829,2.1748252,1.8041958,1.8181818,1.7342657,1.6223776,1.2377622,1.8671329,
                            1.4195804,1.2027972,1.2377622,1.8671329,1.0769231)


#do not modify!!
predictors_sd_ingendep<-c(0.6302054,0.9329961,0.5628087,0.7176169,5.8649321,0.8249173,0.4321375,0.4218731,0.4886176,0.9329961,0.6813206,0.3406242,0.6756592,
                          0.6875087,0.6813206,0.3607067,0.7128293)


predictors<-c(appetite,BDI_changes_sleep,rs62182022,interest_activity_dimension,HRSD_total,SCAN_fatigability,rs1392611,rs10812099,
              rs1891943,HRSD_13,rs9557363,rs151139256,rs28373080,rs7757702,rs76557116,rs11002001,rs2704022)

predictors.std<-(predictors-predictors_mean_ingendep)/predictors_sd_ingendep

logit<-model.coefficients[1]+sum(predictors.std*model.coefficients[2:18])

prob.remission<-exp(logit)/(1+exp(logit))

prob.remission


#Main references:

#BDI (BECK DEPRESSION INVENTORY): A.T. Beck et al. An inventory for measuring depression Arch. General Psychiatry, 4 (1961), pp. 561–571
#HRSD-17 (HAMILTON RATING SCALE FOR DEPRESSION): M. Hamilton. Development of a rating scale for primary depressive illness. Br. J. Soc. Clin. Psychol., 6 (1967), pp. 278–296 17-items version: http://dcf.psychiatry.ufl.edu/files/2011/05/HAMILTON-DEPRESSION.pdf
#SCAN (SCHEDULES FOR CLINICAL ASSESSMENT IN NEUROPSYCHIATRY): J. K. Wing et al. SCAN. Schedules for Clinical Assessment in Neuropsychiatry. Arch.Gen.Psychiatry, 47 (1990), pp. 589-93
#appetite and interest activity dimension, defined as in R. Uher et al. Depression symptom dimensions as predictors of antidepressant treatment outcome: replicable evidence for interest-activity symptoms Psychol. Med., 42 (2012), pp. 967–980

Predicting-remission-Nortriptyline.R

#Predictive model from GENDEP data 
#Antidepressant drug-specific prediction of depression treatment outcomes from genetic and clinical variables.
#Raquel Iniesta raquel.iniesta@kcl.ac.uk

#Use the model to estimate the probability of remission (yes or not) for a patient of depression after 12 weeks of treatment with Nortriptyline. 
#List of baseline predictors that are necessary for the prediction: 

#GENETIC variables: SNP variants coded as 0,1,2 where 0:Homozygous for minor allele; 1:Heterozygous; 2:Homozygous for major allele
#rs6794400 
#rs79693177 
#rs12874087 
#rs2345113 
#rs17091959 
#rs10792321 
#rs199561596 
#rs144829540 
#rs149619279 
#rs34319049 
#rs151132095 
#rs37596 
#rs8053632 
#rs111685823 
#rs4279984 
#rs17057129 
#rs5889536 
#rs34841556 
#rs4773117 
#rs8082631 


#Install the latest version of R from https://cran.r-project.org 

#Copy and paste instructions from STEP 1 and STEP 2 in the R console:

#######STEP 1. Data update.
#Please supply the values for a specific patient. You must replace 0 values by the current patient measurements.
#Then copy and paste in the R console.

rs6794400<-0 
rs79693177<-0  
rs12874087<-0  
rs2345113<-0  
rs17091959<-0  
rs10792321<-0  
rs199561596<-0  
rs144829540<-0  
rs149619279<-0  
rs34319049<-0  
rs151132095<-0  
rs37596<-0  
rs8053632<-0  
rs111685823<-0  
rs4279984<-0  
rs17057129<-0  
rs5889536<-0  
rs34841556<-0  
rs4773117<-0  
rs8082631<-0  

#######STEP 2. Computation of probability of remission 
#Next code lines allow to estimate the probability of remission for an indidivual patient. 
#Please do not modify any value in the code!! Copy and paste in the R console.
#The final computation for the probability of remission will be stored in the object "prob.remission".

#do not modify!!
model.coefficients<-c(-0.034749901,-0.034297552,-0.026953497,-0.027946186,-0.028866071,-0.031566292,-0.032799314,-0.041680584,
                      -0.034297653,-0.025278275,-0.019766966,-0.024609218,-0.024995494,-0.016041970,-0.005547687,-0.005570361,
                      -0.008742997,-0.014872395,0.014553246,0.012370682,0.015627477)

#do not modify!!
predictors_mean_ingendep<-c(1.875912,1.890511,1.445255,1.576642,1.605839,1.839416,1.751825,1.875912,1.693431,1.956204,1.846715,1.627737,1.875912,
                            1.927007,1.941606,1.576642,1.897810,1.087591,1.905109,1.503650)

#do not modify!!
predictors_sd_ingendep<-c(0.3308913,0.3360417,0.6053646,0.5780625,0.5471539,0.4064476,0.4817404,0.3308913,0.4627610,0.2053911,0.4001932,
                          0.5001073,0.3524130,0.3123685,0.2353478,0.5906456,0.3273034,0.7016205,0.3181567,0.6198207)

predictors<-c(rs199561596,rs79693177,rs12874087,rs2345113,rs17091959,rs10792321,rs6794400,rs144829540,rs8053632,rs34319049,
              rs151132095,rs37596,rs149619279,rs111685823,rs4279984,rs17057129,rs5889536,rs34841556,rs4773117,rs8082631 )


predictors.std<-(predictors-predictors_mean_ingendep)/predictors_sd_ingendep

logit<-model.coefficients[1]+sum(predictors.std*model.coefficients[2:21])

prob.remission<-exp(logit)/(1+exp(logit))

prob.remission