rNIX/R/modifyData.R

#' Sum columns in data frame df where some of the entries might be NA
#'
#' @param df data frame
#' @param var1 first column
#' @param var2 second column
#' @param outVar which variable to store sum into
#' @param valIfNA which value to use for NA
#'
#' @return updated data frame
#' @export 
#'
## examples df<-sumWithNA(df,'lesionmtv41','metastasesmtv41','totalmtv41',0)

sumWithNA<-function(df,var1='lesionmtv41',var2='metastasesmtv41',outVar='totalmtv41',valIfNA=0){
   v1=df[,var1]
   v2=df[,var2]
   v1[base::is.na(v1)]=valIfNA
   v2[base::is.na(v2)]=valIfNA
   df[,outVar]=v1+v2
   df
      
}

#' Map NA to value
#'
#' @param df data frame
#' @param var variable to modify
#' @param valIfNA change from NA to this value
#'
#' @return updated data frame
#'
#' @export

mapNA<-function(df,var,valIfNA=0){
   df[base::is.na(df[,var]),var]=valIfNA
   df
}

#' Remove entries with missing variables for selected variables
#'
#' @param x data frame
#' @param vars vector of variable names as strings
#'
#' @return data frame with entries where vars were NA removed
#'
#' @export

remove.NA<-function(x,vars){
   for (v in vars){
      x<-x[!base::is.na(x[,v]),]
   }
   x
}


#' Map variable time OS status to status at cutoff
#'
#' @param df data frame
#' @param cutoff time instance where OS is evaluated, same units as timeVar
#' @param timeVar name of column where times of status evaluation are recorded (from treatment start)
#' @param osVar name of column with status at evaluation (1-alive/progress free, 2-censored, 3-dead/w/disease)
#' @param targetVar name of target variable holding status at cutoff (0-dead, 1-alive, 2-censored)
#'
#' @return updated data frame with targetVar
#' @export 

selectValid<-function(df,cutoff=2,timeVar='years_to_event',osVar='st_osMAP',targetVar='osAtCutoff'){
   #if alive and ytoevent>c -> alive
   #if censored and ytoe>c -> alive
   #if dod and ytovent>c -> alive
   df[,targetVar]=1



   #if alive and ytoe<c -> censored
   df[(df[,timeVar]<cutoff) & df[,osVar] == 1,targetVar]=2

   #if censored and ytoe<c -> censored
   df[df[,timeVar]<cutoff & df[,osVar] == 2,targetVar]=2

   #if dod and ytovent<c -> dead
   df[df[,timeVar]<cutoff & df[,osVar] == 3,targetVar]=0
   df

}


#' Remap/change value of variable according to a lookup table
#'
#' @param df data frame
#' @param origVar name of original variable
#' @param mapVar name of variable with updated values
#' @param valueMap mapping of variables from old to new, named list 
#' where names() are values to be found in data frame and values() contain new values
#' @param num ensure output value is numeric 
#' 
#' @return data frame with updated column
#' @export

remapVariable <- function(df, origVar, mapVar, valueMap, num=TRUE) {
    if (!origVar %in% colnames(df)) {
        base::stop(base::sprintf("Variable '%s' not found in data frame", origVar))
    }
    if (num){
       # Ensures output type consistency
       df[[mapVar]] <- base::vapply(base::as.character(df[[origVar]]), getNewValue, map=valueMap,FUN.VALUE = numeric(1)) 
    }
    else 
       df[[mapVar]] <- base::sapply(df[[origVar]], getNewValue, map=valueMap)
    
    df
}

getNewValue<-function(x,map){
   if (x %in% base::names(map))
      map[[x]]
   else 
      NA
}

#'Set event time
#'
#' @param df data frame
#' @param duration name of the output column
#' @param eventColumn dates of evaluation visits (as.Date)
#' @param startColumn dates of treatment start or diagnosis (as.Date)
#'
#' @return updated data frame
#'
#' @export

setEventTime<-function(df,duration='years_to_event',eventColumn='d_os',startColumn='rtstartdate1'){
   t0=df[,startColumn]
   t1=df[,eventColumn]
   df[,duration] <-base::as.numeric(base::difftime(t1, t0, units = "days")) / 365.25
   df
}

#' Convert string variable to date using prescribed format
#'
#' @param x data frame
#' @param var variable to be converted
#' @param format format of data variable as set by as.Date R function
#'
#' @return data frame with selected variable set as date
#'
#' @export 

convert.to.date<-function(x,var,format="%Y-%m-%d"){
   x[,var]<-base::as.Date(x[,var],format=format)
   x
}