app.R

#App to allow for user to Generate Fake SEND Data for BW, DM, DS, EX, LB, MI,
#TA, TS, and TX Domains
#Requires a SEND nonclinical dataset as a base

#Load Libraries
library(dplyr)
library(bayesplot)
library(haven)
library(Hmisc)
library(MCMCpack)
library(shiny)
library(shinyFiles)
library(shinyWidgets)
library(stringr)
library(this.path)
library(tidyr)
library(tools)
library(ggplot2)

#Set Home Directory
homePath <- dirname(this.path())
setwd(homePath)
#Source Needed Functions
source(paste0(homePath, "/R/fct_Functions.R"))

ui <- fluidPage(

    # Sidebar with user controls for data generation
    sidebarLayout(
        sidebarPanel(
            tags$head(tags$style(type = "text/css", "
              #loadmessage {
                position: relative;
                top: 0px;
                left: 0px;
                width: 100%;
                padding: 5px 0px 5px 0px;
                text-align: center;
                font-weight: bold;
                font-size: 100%;
                color: #000000;
                background-color: #eef66c;
                z-index: 1000;
              }
                ")),
            h2("SEND Sanitizer"),
            tags$hr(),
            h4("1) Choose Example SEND study"),
            #User Inputs SEND study Example
            shinyDirButton("directory", "SEND Example Folder select",
                           "Please select a folder containing SEND .xpt Files"),
            actionButton("submitbutton","Add SEND Folder", class = "btn btn-primary"),
            checkboxGroupInput('chosenfolders','Chosen Folders:'),
            verbatimTextOutput("selected"),
            tags$hr(),
            h4("2) Pick Number of Studies to Generate"),
            #User Inputs number of SEND studies to be generated
            Createme <- sliderInput("Createme", label = "Number of Studies to Generate",
                                    min = 0, max = 50, value = 1, step = 1),
            tags$hr(),
            h4("3) Select from Options"),
            #User has toggle to print generated SEND studies to Folder (T/F)
            tags$p('Print Folder of Generated .XPT File(s) in App Folder'),
            switchInput("PRINT", label = "Print",
                        onLabel = "YES", offLabel = "NO", value = FALSE),

            tags$hr(),
            #User has toggle to include or remove recovery animals (T/F)
            tags$p('Remove Recovery Animals'),
            switchInput("Recovery", label = "Recovery",
                        onLabel = "No", offLabel = "Yes", value = FALSE),
            tags$hr(),
            h4("4) Start Generation Process"),
            #Button to Activate Generation of New SEND Dataset
            actionButton("Start", label = "Generate Datasets")
        ),

        # Blank Main Panel
        mainPanel(
            h3("Generated Data"),
            conditionalPanel(condition = "$('html').hasClass('shiny-busy')",
                             tags$div("Loading...", id="loadmessage")
            ),
            #Display Generated .xpt files as tabsetPanel with each of the generated Domains
            tabsetPanel(
                tabPanel("Body Weight(BW)", dataTableOutput("BW")),
                tabPanel("Demographics (DM)", dataTableOutput("DM")),
                tabPanel("Disposition (DS)", dataTableOutput("DS")),
                tabPanel("Exposure (EX)", dataTableOutput("EX")),
                tabPanel("Laboratory Tests (LB)", dataTableOutput("LB")),
                tabPanel("Microscopic Findings (MI)", dataTableOutput("MI")),
                tabPanel("Trial Arms (TA)", dataTableOutput("TA")),
                tabPanel("Trial Summary (TS)", dataTableOutput("TS")),
                tabPanel("Trial Sets (TX)", dataTableOutput("TX"))
            )
        )
    )
)


server <- function(input, output, session) {

    #Allow for Folder Base Selection of SEND Example Study
    volumes <- c(Home = homePath, "R Installation" = R.home(), getVolumes()())
    shinyDirChoose(input, "directory", roots = volumes, session = session,
                   restrictions = system.file(package = "base"), allowDirCreate = FALSE)
    #Reactive Value to Save FilePaths from shinyDirChoose
    datasetInput <- reactive({
        inputfolders <- c(input$chosenfolders,
                          parseDirPath(volumes, input$directory))
        updateCheckboxGroupInput(session, 'chosenfolders',
                                 choices = unique(inputfolders),
                                 selected = unique(inputfolders))
        print(list('Input folders' = inputfolders))
    })

    #update Checkbox with new Selected Folder
    observeEvent(input$submitbutton,{
        isolate(datasetInput())
    })


    #Server Logic to Generate Fake SEND Data once button is pressed
    observeEvent(input$Start, {

        #Take Options from user inputs
        Createme <- input$Createme
        PRINT <- input$PRINT
        Recovery <- input$Recovery
        ExampleStudies <- datasetInput()
        ExampleStudies$'Input folders' <- unique(ExampleStudies$'Input folders')#Remove Duplicated last Selection
        NumData <- length(ExampleStudies$'Input folders')

        #Make Loop for Loading in the SEND Data per Example Study
        for (i in 1:NumData){
            Name <- paste0('ExampleStudy',as.character(i))
            assign(Name,load.xpt.files(ExampleStudies$`Input folders`[i]))
        }
        Domains <- c("bw","dm","ds","ex","lb","mi","ta","ts","tx")

        #Check that Example Studies are Similar and Consolidate
        if (NumData>1){
            #Generate Names of number of Example Study and concatenate
            Example <- ExampleStudy1
            for (j in 2:NumData){
                Name <- paste0('ExampleStudy',as.character(j))
                #Combine BW, DM, DS, EX, LB, MI, TA, TS, and TX
                Example$bw <- rbind(Example$bw, get(Name)$bw)
                Example$dm <- rbind(Example$dm, get(Name)$dm)
                Example$ds <- rbind(Example$ds, get(Name)$ds)
                Example$ex <- rbind(Example$ex, get(Name)$ex)
                Example$lb <- rbind(Example$lb, get(Name)$lb)
                Example$mi <- rbind(Example$mi, get(Name)$mi)
                Example$ta <- rbind(Example$ta, get(Name)$ta)
                Example$ts <- rbind(Example$ts, get(Name)$ts)
                Example$tx <- rbind(Example$tx, get(Name)$tx)
            }
            #remove unused domains
            Example <- Example[Domains]
            #Check Species are the same
            Species <- getFieldValue(Example$ts, "TSVAL", "TSPARMCD", "SPECIES")
            if (length(unique(Species)) >1){
                stop("ERROR:Species are not the same between SEND Example Studies. Pick one Species.")
            }
            #CHeck Study Type is the same
            SSTYP <- getFieldValue(Example$ts, "TSVAL", "TSPARMCD", "SSTYP")
            if (length(unique(SSTYP)) >1){
                stop("ERROR:Study Types are not the same between SEND Example Studies. Pick one SSTYP.")
            }
            #Check if SEND version is the same
            SNDIGVER <- getFieldValue(Example$ts,"TSVAL", "TSPARMCD", "SNDIGVER")
            if (length(unique(SNDIGVER)) >1){
                stop("ERROR:SEND versions are not the same between SEND Example Studies. Pick one SNDIGVER.")
            }
            #Remove TK SETCDs
            Example$dm <- Example$dm[which(grepl("TK",Example$dm$SETCD) ==FALSE),]
            Example$ta <- Example$ta[which(grepl("Toxicokinetic", Example$ta$ARM) == FALSE),]
            Example$tx <- Example$tx[which(Example$tx$SETCD %in% Example$dm$SETCD),]
            Example$lb <- Example$lb[which(Example$lb$USUBJID %in% Example$dm$USUBJID),]
            Example$mi <- Example$mi[which(Example$mi$USUBJID %in% Example$dm$USUBJID),]
            Example$bw <- Example$bw[which(Example$bw$USUBJID %in% Example$dm$USUBJID),]
            Example$ex <- Example$ex[which(Example$ex$USUBJID %in% Example$dm$USUBJID),]
            Example$ta <- Example$ta[which(Example$ta$ARMCD %in% Example$dm$ARMCD),]

            #Check Dose Levels are equivalent
            if (Recovery == FALSE){
                #Remove Recovery Dose ARMCDs
                doses <- Example$ta[which(grepl("R",Example$ta$ARMCD) == FALSE),c("STUDYID","ARMCD")]
                doses$ARMCD <- as.character(doses$ARMCD)
            }else {
                doses <- Example$ta[,c("STUDYID","ARMCD")]
                doses$ARMCD <- as.character(doses$ARMCD)
            }
            DoseTable <- table(doses)
            if (any(DoseTable == 0)){
                stop("ERROR:ARMCD for Dosing is not equavalent between SEND Example Studies. Try removing Recovery Animals.")
            }

            ##Create ARMCD and DOSE Correlation
            Doses <-Example$ta[,c("ARMCD","ARM")]
            Doses <- Doses[!duplicated(Doses$ARM),]
            Doses$Dose <- NA

        } else {
            #remove unused domains
            Example <- ExampleStudy1 #First study loaded will always be ExampleStudy1
            Example <- Example[Domains]

            #Create ARMCD and DOSE Correlation
            Doses <-data.frame("ARMCD" = as.character(unique(Example$ta$ARMCD)),
                               "Dose" = as.character(unique(Example$ta$ARM)))

        }

        #Get SEND Species, LB TESTCDs and MI Tests from Example Study
        Species <- unique(getFieldValue(Example$ts, "TSVAL", "TSPARMCD", "SPECIES"))
        LBTestCDs <- unique(Example$lb$LBTESTCD)
        MITests <- unique(Example$mi$MISPEC)

        #Find out Animal USUBJIDs for Control and Treated Animals
        if (Recovery == FALSE){
            #Remove Recovery Dose ARMCDs/Doses
            Doses <- Doses[which(grepl("R",Doses$ARMCD) == FALSE),]
        }
        #Replace Dose Levels with Control, LD, MD and HD
        ARMS <- unique(Doses$ARMCD) #Find ARMS levels
        if (Recovery == TRUE){
            #Make MaxDose and Account for "R" doses
            NonrecovArm <- ARMS[which(grepl("R",ARMS) == FALSE)]
            Maxdose <- max(as.numeric(as.character(NonrecovArm)))
            Doses$Dose[which(Doses$ARMCD=="1R")] <- "Control R"
            Doses$Dose[which(Doses$ARMCD=="2R")] <- "LD R"
            Doses$Dose[which(Doses$ARMCD== paste0(Maxdose,"R"))] <- "HD R"
        } else {
            Maxdose <- max(as.numeric(as.character(ARMS)))
        }
        Doses$Dose[which(Doses$ARMCD=="1")] <- "Control"
        Doses$Dose[which(Doses$ARMCD==Maxdose)] <- "HD"
        Doses$Dose[which(Doses$ARMCD=="2")] <- "LD"
        if (length(ARMS) > 3){
            for (i in  3:(max(as.numeric(ARMS)) - 1)){
                j <- i-2
                if (j == 1){
                    Doses$Dose[Doses$ARMCD== as.character(i)] <- "MD"
                } else {
                    Doses$Dose[Doses$ARMCD== as.character(i)] <- paste0("MD", j)
                }
            }
        }
        #Correlate USUBJID with Dose Group
        Subjects <- merge(Example$dm[,c("USUBJID","ARMCD","SEX")], Doses, by = "ARMCD")

        #Double Check Recovery Coding
        RecoveryAnimals <- Example$ds$USUBJID[which(grepl("Recovery",Example$ds$DSTERM) == TRUE)]
        if (Recovery == FALSE){
         if (length(RecoveryAnimals)==0){

         } else if (identical(integer(0),which(Subjects$USUBJID %in% RecoveryAnimals)) == TRUE) {

         }
             else {
             Subjects <- Subjects[-which(Subjects$USUBJID %in% RecoveryAnimals),]
         }
        }
        #Consolidate Severity Methods
        Example$mi$MISEV <- as.character(Example$mi$MISEV)
        Example$mi$MISEV <- str_replace_all(Example$mi$MISEV, "1 OF 5", "1")
        Example$mi$MISEV <- str_replace_all(Example$mi$MISEV, "1 OF 4", "1")
        Example$mi$MISEV <- str_replace_all(Example$mi$MISEV, "PRESENT", "1")
        Example$mi$MISEV <- str_replace_all(Example$mi$MISEV, "MINIMAL", "1")
        Example$mi$MISEV <-  str_replace_all(Example$mi$MISEV, "2 OF 5", "2")
        Example$mi$MISEV <-  str_replace_all(Example$mi$MISEV, "MILD", "2")
        Example$mi$MISEV <-  str_replace_all(Example$mi$MISEV, "3 OF 5", "3")
        Example$mi$MISEV <-  str_replace_all(Example$mi$MISEV, "2 OF 4", "3")
        Example$mi$MISEV <-  str_replace_all(Example$mi$MISEV, "MODERATE", "3")
        Example$mi$MISEV <-  str_replace_all(Example$mi$MISEV, "4 OF 5", "4")
        Example$mi$MISEV <-  str_replace_all(Example$mi$MISEV, "3 OF 4", "4")
        Example$mi$MISEV <-  str_replace_all(Example$mi$MISEV, "MARKED", "4")
        Example$mi$MISEV <-  str_replace_all(Example$mi$MISEV, "5 OF 5", "5")
        Example$mi$MISEV <-  str_replace_all(Example$mi$MISEV, "4 OF 4", "5")
        Example$mi$MISEV <-  str_replace_all(Example$mi$MISEV, "SEVERE", "5")
        Example$mi$MISEV <- replace_na(Example$mi$MISEV, "0")
        Example$mi$MISEV <- ordered(Example$mi$MISEV, levels= c("0","1", "2", "3", "4","5"))


        #Set number of subjects to create based on Example(s)
        SubjectDet <- Subjects %>%
            group_by(Dose) %>%
            count(USUBJID) %>%
            dplyr::mutate(sum = sum(n)) %>%
            dplyr::select(-n)
        SubjectDet <- SubjectDet[!duplicated(SubjectDet$Dose),]
        GeneratedSEND <- list()

        ###############################Generation of Data #############################
        for (j in 1:Createme){
            #Make SENDstudy length of one example study
            onestudy <- as.character(Example$dm$STUDYID[1])
            #Generate base for study to fill with proper SEND format
            SENDstudy <- list( 'dm' = data.frame(Example$dm[which(Example$dm$STUDYID == onestudy),]),
                               'bw' = data.frame(Example$bw[which(Example$bw$STUDYID == onestudy),]),
                               'ds' = data.frame(Example$ds[which(Example$ds$STUDYID == onestudy),]),
                               'ex' = data.frame(Example$ex[which(Example$ex$STUDYID == onestudy),]),
                               'lb' = data.frame(Example$lb[which(Example$lb$STUDYID == onestudy),]),
                               'mi' = data.frame(Example$mi[which(Example$mi$STUDYID == onestudy),]),
                               'ta' = data.frame(Example$ta[which(Example$ta$STUDYID == onestudy),]),
                               'ts' = data.frame(Example$ts[which(Example$ts$STUDYID == onestudy),]),
                               'tx' = data.frame(Example$tx[which(Example$tx$STUDYID == onestudy),]))

            #Create StudyID and Compound Name for generated study
            studyID <- floor(runif(1, min = 10000, max = 100000))
            Compound <- paste0("Fake-Drug ", floor(runif(1, min = 1, max = 100000)))

            #Generate TS Data
            #Keeps: Study design, GLP flag and type, duration, species, age, vehicle, dosing duration
            #Replaces: dates, study title, study facility, study compound, primary treatment
            #Removes: Study Director, Animal Purchasing Location, and Test Facility Country

            #Replace StudyID
            SENDstudy$ts$STUDYID <- rep(studyID, nrow(SENDstudy$ts))

            #Find Date TSPARMCDs and replace
            daterows <- grep("DTC", SENDstudy$ts$TSPARMCD)
            SENDstudy$ts[daterows, "TSVAL"] <- rep("XXXX-XX-XX",length(daterows))

            #Replace Study Facility Name and Location
            rows <- grep("TSTF", SENDstudy$ts$TSPARMCD)
            SENDstudy$ts[rows, "TSVAL"] <- rep("FAKE FACILITY", length(rows))

            #Replace Study Compound/Primary Treatment CAS number, name, and unique ingredient ID
            rows <- grep("TRT",SENDstudy$ts$TSPARMCD)
            SENDstudy$ts[rows, "TSVAL"] <- rep(Compound, length(rows))
            if (NumData > 1){
                Vehicles <- Example$ts[grep("TRTV",Example$ts$TSPARMCD),"TSVAL"]
                #Remove any N/A or "NOT AVAILABLE"
                Vehicles <- Vehicles[which(str_detect(Vehicles,"NOT AVAILABLE") == FALSE)]
                Vehicles <- Vehicles[which(str_detect(Vehicles,"NA") == FALSE)]
                #Check if values are the same for vehicle and concatinate if not
                if (length(unique(Vehicles)) == 1){
                    SENDstudy$ts[grep("TRTV",SENDstudy$ts$TSPARMCD),"TSVAL"] <- Vehicles
                } else {
                    Vehicles <- paste(Vehicles, collapse = " / ")
                    SENDstudy$ts[grep("TRTV",SENDstudy$ts$TSPARMCD),"TSVAL"] <- Vehicles
                }
            } else {
                SENDstudy$ts[grep("TRTV",SENDstudy$ts$TSPARMCD),"TSVAL"] <- Example$ts[grep("TRTV",Example$ts$TSPARMCD),"TSVAL"]
            }

            #Replace Study Title
            rows <- grep("STITLE", SENDstudy$ts$TSPARMCD)
            duration <- getFieldValue(SENDstudy$ts, "TSVAL", "TSPARMCD", "DOSDUR")
            SENDstudy$ts[rows, "TSVAL"] <- paste0(Compound, ": A ",  duration," Fake Study in ",
                                                  Species)

            #Clean up Vehicle
            idx <- which(grepl("TRTV",SENDstudy$ts$TSPARMCD)==TRUE)
            SENDstudy$ts[idx,"TSVAL"] <- paste0("VEHICLE")

            #Remove Identifying Information
            RemoveTerms <- c("TFCNTRY","STDIR","SPLRNAM","TFCNTRY","TRMSAC","SSPONSOR","SPREFID", "SPLRLOC",
                             "PINV","STMON","TSLOC","TSCNTRY","DIET","WATER", "PCLASS")
            for (term in RemoveTerms){
                #Check index for Term
                idx <- which(SENDstudy$ts$TSPARMCD == term)
                #Remove Term
                SENDstudy$ts$TSVAL[idx] <- ""
            }


            #Generate TA Data
            #Keeps: EPOCH, ELEMEND, ETCD, TAETORD, DOMAIN
            #Replaces: StudyID, ARM

            #Replace StudyID
            SENDstudy$ta$STUDYID <- rep(studyID, nrow(SENDstudy$ta))

            #Replace ARM
            SENDstudy$ta$ARM <- as.character(SENDstudy$ta$ARM)
            if (Recovery == FALSE){
                SENDstudy$ta <- SENDstudy$ta[which(grepl("R",SENDstudy$ta$ARMCD) == FALSE),]
            }
            for (arms in unique(as.character(SENDstudy$ta$ARMCD))){
                row <- which(arms == Doses$ARMCD)
                Dosein <- unique(Doses$Dose[row])
                rows <- which(arms == SENDstudy$ta$ARMCD)
                SENDstudy$ta[rows,"ARM"] <- rep(Dosein, length(rows))
            }

            #Generate DM Data
            #Keeps: Number of each gender animals in each treatment group
            #Replaces: StudyID, USUBJID, Dates

            #Account for discrepancies possible in dm with recovery
            SENDstudy$dm <- SENDstudy$dm[which(SENDstudy$dm$USUBJID %in% Subjects$USUBJID),]
            #Find number of subjects in each group of each gender
            ControlAnimals <- SENDstudy$dm[which(SENDstudy$dm$ARMCD == 1),]
            Gendersplit <- table(ControlAnimals$SEX)

            #Replace StudyID
            SENDstudy$dm$STUDYID <- rep(studyID, nrow(SENDstudy$dm))

            #Generate new USUBJIDs using SBJID
            NEWUSUBJID <- paste0(studyID, "-" ,SENDstudy$dm$SUBJID)
            USUBJIDTable <- data.frame(USUBJID = SENDstudy$dm$USUBJID,
                                       NEWUSUBJID = NEWUSUBJID)
            SENDstudy$dm$USUBJID <- NEWUSUBJID

            #Replace Dates
            cols <- grep("DTC", colnames(SENDstudy$dm))
            SENDstudy$dm[,cols] <- rep("XXXX-XX-XX",length(SENDstudy$dm$STUDYID))

            #Replace ARM
            SENDstudy$dm$ARM <- as.character(SENDstudy$dm$ARM)
            if (Recovery == FALSE){
                SENDstudy$dm <- SENDstudy$dm[which(grepl("R",SENDstudy$dm$ARMCD) == FALSE),]
            }
            for (arms in unique(SENDstudy$dm$ARMCD)){
                row <- which(arms == Doses$ARMCD)
                Dosein <- unique(Doses$Dose[row])
                rows <- which(arms == SENDstudy$dm$ARMCD)
                SENDstudy$dm[rows,"ARM"] <- rep(Dosein, length(rows))
            }
            ExampleSubjects <- SENDstudy$dm[,c("USUBJID", "ARM","SUBJID","SEX")]

            #Make Factors Characters for Correct .xpt creation
            SENDstudy$dm$SEX <- as.character(SENDstudy$dm$SEX)
            SENDstudy$dm$AGEU <- as.character(SENDstudy$dm$AGEU)

            #Generate DS Data
            #Keeps: VISITDY
            #Replaces: StudyID, USUBJID, USUBJID, Dates

            #Account for discrepancies possible in ds
            SENDstudy$ds <- SENDstudy$ds[which(SENDstudy$ds$USUBJID %in% Subjects$USUBJID),]

            #Account for VISITDY for Recovery Animals
            if (Recovery == FALSE){
                SENDstudy$ds <- SENDstudy$ds[which(grepl("Recovery",SENDstudy$ds$DSTERM) == FALSE),]
            }

            #ADD Generated USUJIDs and terminal VISITDY to new DS
            SENDstudy$ds$USUBJID <- SENDstudy$dm$USUBJID

            #Calculate percentage of Terminal Sacrifice in each ARM
            TerminalSac <- merge(Example$ds, Subjects, by = "USUBJID")
            TerminalSac <- TerminalSac %>%
                group_by(Dose) %>%
                count(DSDECOD) %>%
                dplyr::mutate(percent = n/sum(n)) %>%
                dplyr::select(-n)

            #Replace StudyID
            SENDstudy$ds$STUDYID <- rep(studyID, nrow(SENDstudy$ds))

            #Replace Dates
            cols <- grep("DTC", colnames(SENDstudy$ds))
            SENDstudy$ds[,cols] <- rep("XXXX-XX-XX",length(SENDstudy$ds$STUDYID))

            #Make Generated DS Terminal Sacrifice percentage match expectation
            for(Dose in unique(Doses$Dose)){
                Percen <- TerminalSac[which(TerminalSac$Dose == Dose & TerminalSac$DSDECOD == 'TERMINAL SACRIFICE'), 'percent']
                Subjs <- ExampleSubjects$USUBJID[which(ExampleSubjects$ARM == Dose)]
                Gendata <- sample(c("TERMINAL SACRIFICE", "FOUND DEAD"),length(Subjs), replace = TRUE, prob = c(Percen, (1-Percen)))
                #Add in Randomized Data to Subjects in Each Group
                SENDstudy$ds[which(SENDstudy$ds$USUBJID %in% Subjs), "DSDECOD"] <- Gendata
                #Make VISITDY Appropriate
                VISTDY <- max(SENDstudy$ds[which(SENDstudy$ds$USUBJID %in% Subjs), "VISITDY"], na.rm = TRUE)
                SENDstudy$ds[which(SENDstudy$ds$USUBJID %in% Subjs), "VISITDY"] <- VISTDY
            }
            #Ensure DSTERM/VISITDY is APPROPRIATE
            SENDstudy$ds <-SENDstudy$ds %>%
                mutate(DSTERM = ifelse(DSDECOD == 'FOUND DEAD','Found Dead','Terminal necropsy')) %>%
                mutate(VISITDY = ifelse(DSDECOD == 'FOUND DEAD',NA,VISITDY))

            #Generate TX data
            #Keeps: SETCD
            #Replaces: SET

            #Replace StudyID
            SENDstudy$tx$STUDYID <- rep(studyID, nrow(SENDstudy$tx))

            #Account for Recovery Animals
            if (Recovery == FALSE){
                SENDstudy$tx <- SENDstudy$tx[which(grepl("R",SENDstudy$tx$SETCD) == FALSE),]
            }

            #Replace SET with Blinded Notation
            ARMS <- getFieldValue(SENDstudy$tx,'TXVAL','TXPARM', 'Arm Code')
            SETS <- SENDstudy$tx$SET[which(SENDstudy$tx$TXPARM == 'Arm Code')]
            ARMtoset <- data.frame('Arm' = ARMS, 'Set' = SETS)
            SENDstudy$tx$TXVAL <- as.character(SENDstudy$tx$TXVAL)
            SENDstudy$tx$SET <- as.character(SENDstudy$tx$SET)
            for (i in 1:nrow(ARMtoset)){
                ARMCD <- as.character(ARMtoset$Arm[i])
                SETCD <- as.character(ARMtoset$Set[i])
                DoseCover <- Doses$Dose[which(Doses$ARMCD == ARMCD)]
                if(identical(DoseCover, character(0)) == TRUE){
                DoseCover <- ""
                }
                SENDstudy$tx$TXVAL[which(grepl(SETCD,SENDstudy$tx$TXVAL)== TRUE)] <- unique(DoseCover)
                SENDstudy$tx$SET[which(SENDstudy$tx$SET == SETCD)] <- unique(DoseCover)
            }
            #Replace Factors with Characters
            SENDstudy$tx$SETCD <- as.character(SENDstudy$tx$SETCD)
            SENDstudy$tx$TXPARM <- as.character(SENDstudy$tx$TXPARM)
            SENDstudy$tx$TXPARMCD <- as.character(SENDstudy$tx$TXPARMCD)

            #Remove Identifying Group Names
            idx <- which(grepl("GRPLBL",SENDstudy$tx$TXPARMCD) == TRUE)
            SENDstudy$tx$TXVAL[idx] <- paste0("GROUP: ", SENDstudy$tx$SET[idx])

            #Generate EX data
            #Keeps: ESDOSFRM, EXDOSFRQ, EXROUTE, EXTRTV, EXSTDY, EXDOSEU, EXVATMU, EXSTDY
            #Replaces: EXTRT, EXLOT, EXSTDTC, EXVAMT, EXDOSE, USUBJID, STUDYID

            #ADD Generated USUBJID, EXTRT name, and STUDYID
            SENDstudy$ex$STUDYID <- rep(studyID, nrow(SENDstudy$ex))
            SENDstudy$ex$EXTRT <- rep(Compound, length(SENDstudy$ex$EXTRT))
            #Remove Recovery if Needed
            if (Recovery == FALSE){
                NonRecovSub <- Example$dm$USUBJID[which(grepl("R",Example$dm$ARMCD) == FALSE)]
                SENDstudy$ex <- SENDstudy$ex[which(SENDstudy$ex$USUBJID %in% NonRecovSub),]
            }
            SENDstudy$ex$USUBJID <- as.character(SENDstudy$ex$USUBJID)
            SENDstudy$ex <- merge( USUBJIDTable,SENDstudy$ex, by = "USUBJID")
            SENDstudy$ex <- SENDstudy$ex[,!(names(SENDstudy$ex) %in% "USUBJID")]
            names(SENDstudy$ex)[names(SENDstudy$ex) == "NEWUSUBJID"] <- "USUBJID"

            #Generate Fake EXLOT
            Lotnum <- length(levels(SENDstudy$ex$EXLOT))
            Lot <- paste0("Fake", floor(runif(Lotnum, min = 1, max = 100000)))
            for (i in 1:Lotnum){
                levels(SENDstudy$ex$EXLOT)[i] <-Lot[i]
            }

            #Remove Dates
            cols <- grep("DTC", colnames(SENDstudy$ex))
            SENDstudy$ex[,cols] <- rep("XXXX-XX-XX",length(SENDstudy$ex$STUDYID))

            #Find Distribution of Dose to Vehicle (EXDOSE) to (EXVAMT)
            Dist <- SENDstudy$ex$EXDOSE/SENDstudy$ex$EXVAMT
            #Generate EXDOSE and EXVAMT Numbers based on expectation
            Gen <- round(runif(length(Dist),min=min(SENDstudy$ex$EXVAMT), max = max(SENDstudy$ex$EXVAMT)),2)
            SENDstudy$ex$EXVAMT <- Gen
            SENDstudy$ex$EXDOSE <- Gen*Dist

            #Make Factors as Characters
            SENDstudy$ex$EXDOSU <- as.character(SENDstudy$ex$EXDOSU)
            SENDstudy$ex$EXROUTE <- as.character(SENDstudy$ex$EXROUTE)
            SENDstudy$ex$EXVAMTU <- as.character(SENDstudy$ex$EXVAMTU)
            SENDstudy$ex$EXTRTV <- as.character(SENDstudy$ex$EXTRTV)
            SENDstudy$ex$EXDOSFRQ <- as.character(SENDstudy$ex$EXDOSFRQ)
            SENDstudy$ex$EXDOSFRM <- as.character(SENDstudy$ex$EXDOSFRM)

            #Remove Identifying Terms
            RemoveTerms <- c("SPLRNAM","SSPONSOR","SPREFID", "SPLRLOC")
            for (term in RemoveTerms){
                #Check index for Term
                idx <- which(SENDstudy$tx$TXPARMCD == term)
                #Remove Term
                SENDstudy$tx$TXVAL[idx] <- ""
            }


            #Generates BW Data
            #Keeps: BWORRESU, BWTESTCD, BWSTRESU
            #Replaces: STUDYID, USUBJID, BWORRES, BWSTRESC, BWSTRESN, and BWDTC
            #Removes: BWBLFL

            #Account for TK discrepancies possible in BW
            SENDstudy$bw <- SENDstudy$bw[which(SENDstudy$bw$USUBJID %in% Subjects$USUBJID),]

            #Add Generated StudyID and USUBJID
            SENDstudy$bw$STUDYID <- rep(studyID, nrow(SENDstudy$bw))
            SENDstudy$bw$USUBJID <- as.character(SENDstudy$bw$USUBJID)
            SENDstudy$bw <- merge( USUBJIDTable,SENDstudy$bw, by = "USUBJID")
            SENDstudy$bw <- SENDstudy$bw[,!(names(SENDstudy$bw) %in% "USUBJID")]
            names(SENDstudy$bw)[names(SENDstudy$bw) == "NEWUSUBJID"] <- "USUBJID"

            #Remove Dates
            cols <- grep("DTC", colnames(SENDstudy$bw))
            SENDstudy$bw[,cols] <- rep("XXXX-XX-XX",length(SENDstudy$bw$STUDYID))

            #Remove BWBLFL
            SENDstudy$bw$BWBLFL <- NA

            #Find average weight behavior by dose and gender in Example
            BWFindings <- merge(Subjects, Example$bw[,c("USUBJID", "BWTESTCD", "BWSTRESN","BWDY")], by = "USUBJID")
            BWSummary <- BWFindings %>%
                group_by(Dose, BWTESTCD,BWDY,SEX) %>%
                mutate(ARMavg = mean(BWSTRESN, na.rm = TRUE)) %>%
                mutate(ARMstdev = sd(BWSTRESN,na.rm = TRUE))

            #Make Model of weight using MCMCregress
            for (Dose in unique(Doses$Dose)){
                for (gender in unique(ExampleSubjects$SEX)){
                    #Limit to proper gender subjects
                    Subjs <- ExampleSubjects$USUBJID[which(ExampleSubjects$ARM == Dose & ExampleSubjects$SEX == gender)]
                    Sub <- Subjects$USUBJID[which(Subjects$Dose == Dose & Subjects$SEX == gender)]
                    SubBWFindings <- BWFindings[which(BWFindings$USUBJID %in% Sub),]
                    line <- data.frame(BWSTRESN = SubBWFindings$BWSTRESN, Day= SubBWFindings$BWDY, Dose = SubBWFindings$Dose)
                    #Make model of weight over time per dose group
                    if (Species %in% c("DOG", "MONKEY")){
                        #Linear fit
                        posterior <- MCMCregress(BWSTRESN~Day, b0=0, B0 = 0.1, data = line)
                    } else {
                        #Log fit
                        posterior <- MCMCregress(log(BWSTRESN)~Day, b0=0, B0 = 0.1, data = line)
                    }
                    #Sample model to fill in Example using Subjs
                    #Per individual, sample from postieror and derive line for their response
                    Fit <- sample(1:nrow(posterior), size=length(Subjs))
                    sn <-1
                    #Use that fit to generate new animal data
                    for (Subj in Subjs){
                        BWDYs <-SENDstudy$bw$BWDY[which(SENDstudy$bw$USUBJID == Subj)]
                        SubFit <- Fit[sn]
                        #Calculate BWSTRESN per BWDY in Model
                        if (Species %in% c("DOG", "MONKEY")){
                            #Linear fit
                            GenerData <- data.frame(BWSTRESN = posterior[SubFit,1]+posterior[SubFit,2]*BWDYs,
                                                    BWDY = BWDYs)
                        } else {
                            #Log Fit
                            GenerData <- data.frame(BWSTRESN = exp(posterior[SubFit,1]+posterior[SubFit,2]*BWDYs),
                                                    BWDY = BWDYs)
                        }
                        #Add in noise to fit
                        stdev <- unique(BWSummary$ARMstdev[which(BWSummary$Dose == Dose & BWSummary$SEX == gender)])
                        GenerData$BWSTRESN <- GenerData$BWSTRESN + rnorm(length(GenerData$BWSTRESN), mean = 0, sd = (stdev/2))
                        #Fill into SENDstudy being generated
                        for (day in BWDYs){
                            idx <-which(SENDstudy$bw$USUBJID %in% Subj &
                                            SENDstudy$bw$BWDY %in% day)
                            idx2 <- which(GenerData$BWDY %in% day)
                            SENDstudy$bw$BWSTRESN[idx] <- round(GenerData$BWSTRESN[idx2],2)
                        }
                        sn <- sn+1
                    }
                }
            }

            #Testing code to graph fit compared to average of that ARM
            #Convert all BW into data.frame
            Subjs <- ExampleSubjects$USUBJID[which(ExampleSubjects$ARM == "HD" & ExampleSubjects$SEX == "M")]
             TEST <- SENDstudy$bw[which(SENDstudy$bw$USUBJID %in% Subjs), c("BWDY","BWSTRESN","USUBJID")]
            #Add Average of BWSummary
              TEST <- merge(TEST, unique(BWSummary[which(BWSummary$Dose == "HD" & BWSummary$SEX == "M"), c("BWDY","ARMavg")]), by = c("BWDY"))
              p <- ggplot(data= TEST, aes(x=BWDY,y = BWSTRESN, group=USUBJID, color = "Simulated Animal Data"))+ geom_line()+
                    geom_line(aes(y = ARMavg, label="Average of Source Data", color = "Average of Source Data")) +
                    ggtitle("HD M Weight Distribution Comparison")+
                     labs(x='BWDY (Days)', y="Weight") + scale_color_manual(name = "Legend",
                                                                            values = c("darkred","steelblue"),
                                                                            breaks = c("Simulated Animal Data",
                                                                                       "Average of Source Data"))
              print(p)

            #Generate New Data Based on Example >>> rnorm method option, replaces MCMCregress with more averaged values
            # for (Dose in unique(Doses$Dose)){
            #     for (gender in unique(ExampleSubjects$SEX)){
            #         Subjs <- ExampleSubjects$USUBJID[which(ExampleSubjects$ARM == Dose & ExampleSubjects$SEX == gender)]
            #         Sub <- Subjects$USUBJID[which(Subjects$Dose == Dose & Subjects$SEX == gender)]
            #         GroupTests <- SENDstudy$bw$BWTESTCD[which(SENDstudy$bw$USUBJID %in% Subjs)]
            #         for (Test in unique(GroupTests)){
            #             Days <- unique(BWSummary$BWDY[which(BWSummary$USUBJID %in% Sub & BWSummary$BWTESTCD %in% Test)])
            #             for (day in Days){
            #                 #make indexs
            #                 idx <-which(SENDstudy$bw$USUBJID %in% Subjs & SENDstudy$bw$BWTESTCD %in% Test &
            #                                 SENDstudy$bw$BWDY %in% day)
            #                 idxs <-which(BWSummary$USUBJID %in% Sub & BWSummary$BWTESTCD %in% Test &
            #                                  BWSummary$BWDY %in% day)
            #                 #Filter Group mean and stdev
            #                 Testavg <- unique(BWSummary$ARMavg[idxs])
            #                 Teststdev <- unique(BWSummary$ARMstdev[idxs])
            #                 #use rnorm() to generate data
            #                 GenerData <- suppressWarnings(round(rnorm(n=length(idx), mean = Testavg, sd = Teststdev),2))
            #                 #Fill tests properly with created values
            #                 SENDstudy$bw$BWSTRESN[idx] <- GenerData
            #             }
            #         }
            #     }
            # }

            #Make BWSTRESC and BWORRES match
            SENDstudy$bw$BWSTRESC <- as.character(SENDstudy$bw$BWSTRESN)
            SENDstudy$bw$BWORRES <- SENDstudy$bw$BWSTRESN
            SENDstudy$bw$BWORRESU <- SENDstudy$bw$BWSTRESU

            #Make Factors as Characters
            SENDstudy$bw$BWORRESU <- as.character(SENDstudy$bw$BWORRESU)
            SENDstudy$bw$BWSTRESU <- as.character(SENDstudy$bw$BWSTRESU)
            SENDstudy$bw$BWTESTCD <- as.character(SENDstudy$bw$BWTESTCD)
            SENDstudy$bw$BWTEST <- as.character(SENDstudy$bw$BWTEST)
            SENDstudy$bw$BWDY <- as.character(SENDstudy$bw$BWDY)
            if (any(grepl('VISITDY',colnames(SENDstudy$bw)) == TRUE)){
                SENDstudy$bw$VISITDY <- as.character(SENDstudy$bw$VISITDY)
            }

            #Generates NUMERICAL LB Data
            #Keeps: DOMAIN, LBTESTs, LBTESTCD, LBDY, LBDY, LBCAT
            #Replaces: STUDYID, USUBJID, LBORRES, LBSTRESC, LBSTRESN, and LBDTC
            #Removes: LBBLFL

            #Account for TK discrepancies possible in LB
            SENDstudy$lb <- SENDstudy$lb[which(SENDstudy$lb$USUBJID %in% Subjects$USUBJID),]

            #Replace StudyID and USUBJID
            SENDstudy$lb$STUDYID <- rep(studyID, nrow(SENDstudy$lb))
            SENDstudy$lb <- merge( USUBJIDTable,SENDstudy$lb, by = "USUBJID")
            SENDstudy$lb <- SENDstudy$lb[,!(names(SENDstudy$lb) %in% "USUBJID")]
            names(SENDstudy$lb)[names(SENDstudy$lb) == "NEWUSUBJID"] <- "USUBJID"

            #Remove Dates
            cols <- grep("DTC", colnames(SENDstudy$lb))
            SENDstudy$lb[,cols] <- rep("XXXX-XX-XX",length(SENDstudy$lb$STUDYID))
            #Find out value range per treatment group
            LBFindings <- merge(Subjects, Example$lb[,c("USUBJID", "LBTESTCD", "LBSPEC", "LBSTRESN","LBDY")], by = "USUBJID")
            LBSummary <- LBFindings %>%
                group_by(Dose, LBTESTCD,LBDY,SEX) %>%
                mutate(ARMavg = mean(LBSTRESN, na.rm = TRUE)) %>%
                mutate(ARMstdev = sd(LBSTRESN,na.rm = TRUE))
            #Remove incomplete StudyTests
            LBSummary <- na.omit(LBSummary)
            SENDstudy$lb <- SENDstudy$lb[which(SENDstudy$lb$LBTESTCD %in% LBSummary$LBTESTCD),]
            SENDstudy$lb <- SENDstudy$lb[which(SENDstudy$lb$LBSPEC %in% c('WHOLE BLOOD', 'SERUM', 'URINE')),]

            #Create a distribution of values using MCMC for LBSTRESN
            for (Dose in unique(Doses$Dose)){
                for (gender in unique(ExampleSubjects$SEX)){
                    print(paste0(Dose, " - ", gender))
                    Subjs <- ExampleSubjects$USUBJID[which(ExampleSubjects$ARM == Dose & ExampleSubjects$SEX == gender)]
                    Sub <- Subjects$USUBJID[which(Subjects$Dose == Dose & Subjects$SEX == gender)]
                    GroupTests <- SENDstudy$lb[which(SENDstudy$lb$USUBJID %in% Subjs), c("LBTESTCD","LBSPEC")]
                    for (lbspec in unique(GroupTests$LBSPEC)){
                        Days <- unique(LBSummary$LBDY[which(LBSummary$USUBJID %in% Sub & LBSummary$LBSPEC %in% lbspec)])
                        LBDATAs <- LBSummary[which(LBSummary$LBSPEC %in% lbspec),]
                        LBDATAs <- LBDATAs[which(LBDATAs$USUBJID %in% Sub),]
                        #Remove Tests that have a ARMstev of 0 (meaning they likely don't have enough data)
                        LBDATAs <- LBDATAs[which(LBDATAs$ARMstdev != 0),]
                        #Remove tests that do not have enough data (i.e. all days)
                        Testspread <- table(droplevels(LBDATAs$LBTESTCD), LBDATAs$LBDY)
                        rowsub <- apply(Testspread,1, function(row) all(row !=0))
                        highDataTests <- rownames(Testspread)[rowsub]
                        if (length(highDataTests) <= 1){
                           ToRemove <- which(GroupTests$LBSPEC %in% c(lbspec))
                           SENDstudy$lb <- SENDstudy$lb[-ToRemove,]
                           next #Too little data; SKips loop
                        }
                        LBDATAs <- LBDATAs[which(LBDATAs$LBTESTCD %in% highDataTests),]
                        # how to make line with varying amount of variables
                        line <- data.frame(USUBJID= LBDATAs$USUBJID, LBSTRESN = LBDATAs$LBSTRESN, Day= LBDATAs$LBDY, LBTEST = LBDATAs$LBTESTCD)
                        line <- distinct(line) #check for and remove duplicate rows
                        line <- reshape(line, idvar = c("USUBJID","Day"), timevar = 'LBTEST', direction = "wide")
                        line <- sapply(line[,2:ncol(line)], as.numeric)
                        colnames(line) <- gsub("LBSTRESN.","",colnames(line))
                        line <- as.data.frame(line)
                        #Remove NA values (fit cannot have them)
                        line <- na.omit(line)
                        for (test in unique(LBDATAs$LBTESTCD)){
                            Vars <- setdiff(colnames(line),c("Day",test))
                            if (length(Vars) > 10){ #limit Vars to 2 random variables for computation time
                                Vars <- sample(Vars, 2)
                            }
                            #Repeating fit PER test with interaction from other tests in that lbspec
                            equation <- paste0(Vars, sep= '*Day',collapse = " + ")
                            #Error test
                            print(paste0(lbspec, " - ", test ))
                            #Make Fit
                            LBfit <- MCMCpack::MCMCregress(as.formula(paste0(test, " ~ ",equation)), b0=0, B0 = 0.1, data = line)
                            #Sample Model 'Per Individual animal'
                            Fit <- sample(1:nrow(LBfit), size=length(Subjs))
                            sn <-1
                            for (Subj in Subjs) {
                                LBFit <- LBfit[Fit[sn],]
                                #Make LBSTRESN Fit for that variable
                                DayVars <- which(grepl("Day",names(LBFit)) == TRUE) #Find break between interaction variables and other variables
                                InteractionVars <- tail(DayVars,length(DayVars)-1) #Original Day Variable will be first found
                                #Make Equation Based on Varying length of Variables
                                #LBFit[1] is always the intercept
                                LBTESTVAR <- LBFit[1]
                                #Then it will be individual Variable coeff*their variables (including Day)
                                for (num in 2:(InteractionVars[1]-1)){
                                    testnm <- names(LBFit[num])
                                    LBTESTVAR <- LBTESTVAR + LBFit[num]*line[,testnm]
                                }
                                #Then interaction variables coeff * their variables will be added
                                for (num2 in InteractionVars){
                                  testnm <- unlist(strsplit(names(LBFit[num2]),":"))[2]
                                  LBTESTVAR <- LBTESTVAR + LBFit[num2]*line[,"Day"]*line[,testnm]
                                }
                                #Add Variance using stdev/rnorm
                                stdev <- unique(LBSummary[which(LBSummary$Dose == Dose & LBSummary$SEX == gender & LBSummary$LBTESTCD == test),c('ARMstdev','LBDY')])
                                LBTESTVAR <- abs(LBTESTVAR + rnorm(length(LBTESTVAR), mean = 0, sd = (stdev$ARMstdev)))

                                #Fill DataFrame to allocate to fake individual based on Day once variance is added
                                GenerLBData <- data.frame(LBSTRESN = 0,
                                                          LBDy = 0,
                                                          LBTESTCD = test)
                                avrgs <- unique(LBSummary[which(LBSummary$Dose == Dose & LBSummary$SEX == gender & LBSummary$LBTESTCD == test),c('ARMavg','LBDY')])
                                #rein in values to the days
                                for (Dayz in Days){
                                    Val <- LBTESTVAR[which.min(abs(LBTESTVAR -avrgs$ARMavg[which(avrgs$LBDY == Dayz)]))]
                                    GenerLBData[nrow(GenerLBData)+1,] <- c(Val ,Dayz, test)
                                }
                                GenerLBData <- GenerLBData[2:nrow(GenerLBData),]
                                # For each day store the value in generated dataset SENDstudy
                                for (day in Days){
                                    idx <-which(SENDstudy$lb$USUBJID %in% Subj & SENDstudy$lb$LBTESTCD %in% test &
                                                    SENDstudy$lb$LBDY %in% day)
                                    idx2 <- which(GenerLBData$LBDy %in% day)
                                    SENDstudy$lb$LBSTRESN[idx] <- round(as.numeric(GenerLBData$LBSTRESN[idx2]),3)
                                }
                                #add to subject count before new subject done
                                sn <- sn+1
                            }
                        }
                    }
                }
            }

            #Testing code to graph fit compared to average of that ARM (HD, M, Selected LB tests)
            Subjs <- ExampleSubjects$USUBJID[which(ExampleSubjects$ARM == "HD" & ExampleSubjects$SEX == "M")]
            TEST <- SENDstudy$lb[which(SENDstudy$lb$USUBJID %in% Subjs), c("LBDY","LBSTRESN","LBTESTCD","USUBJID")]

            for (SampleTests in c("RBC","ALB","SPGRAV") ){
                #Add Average of LBSummary
                TEST2 <- TEST[which(TEST$LBTESTCD %in% SampleTests),]
                TEST2 <- merge(TEST2, unique(LBSummary[which(LBSummary$Dose == "HD" & LBSummary$LBTESTCD == SampleTests & LBSummary$SEX == "M"), c("LBDY","ARMavg")]), by = c("LBDY"))
                #Make plot per test
                p <- ggplot(data= TEST2, aes(x=factor(LBDY),y = LBSTRESN, group=USUBJID, color = "Simulated Animal Data"))+ geom_line()+
                    geom_point(aes(x=factor(LBDY),y = LBSTRESN, group=USUBJID, color = "Simulated Animal Data"))+
                    geom_line(aes(y = ARMavg, label="Average of Source Data", color = "Average of Source Data")) +
                    geom_point(aes(y = ARMavg, label="Average of Source Data", color = "Average of Source Data"))+
                    scale_x_discrete()+
                    ggtitle(paste0("HD M Distribution Comparison ", SampleTests))+
                    labs(x='LBDY (Days)', y=paste0(SampleTests, " Values")) + scale_color_manual(name = "Legend",
                                                                           values = c("darkred","steelblue"),
                                                                           breaks = c("Simulated Animal Data",
                                                                                      "Average of Source Data"))
                print(p)
            }

            # #Create distributions of values for LBSTRESN >>> rnorm method option, replaces MCMCregress with more averaged values
            # for (Dose in unique(Doses$Dose)){
            #     for (gender in unique(ExampleSubjects$SEX)){
            #         Subjs <- ExampleSubjects$USUBJID[which(ExampleSubjects$ARM == Dose & ExampleSubjects$SEX == gender)]
            #         Sub <- Subjects$USUBJID[which(Subjects$Dose == Dose & Subjects$SEX == gender)]
            #         GroupTests <- SENDstudy$lb$LBTESTCD[which(SENDstudy$lb$USUBJID %in% Subjs)]
            #         for (lbtest in unique(GroupTests)){
            #             Days <- unique(LBSummary$LBDY[which(LBSummary$USUBJID %in% Sub & LBSummary$LBTESTCD %in% lbtest)])
            #             for (day in Days){
            #                 idx <-which(SENDstudy$lb$USUBJID %in% Subjs & SENDstudy$lb$LBTESTCD %in% lbtest &
            #                                 SENDstudy$lb$LBDY %in% day)
            #                 idxs <-which(LBSummary$USUBJID %in% Sub & LBSummary$LBTESTCD %in% lbtest &
            #                                  LBSummary$LBDY %in% day)
            #                 #Filter Group mean and stdev
            #                 Testavg <- unique(LBSummary$ARMavg[idxs])
            #                 Teststdev <- unique(LBSummary$ARMstdev[idxs])
            #                 #use rnorm to generate new test values based around normal distribution
            #                 GenerData <- suppressWarnings(round(rnorm(n=length(idx), mean = Testavg, sd = Teststdev),2))
            #                 #Fill tests properly with created values
            #                 SENDstudy$lb$LBSTRESN[idx] <- GenerData
            #             }
            #         }
            #     }
            # }

            #Plot to look at correlation between Urine SPGRAV and Urine Volume
            Subjs <- ExampleSubjects$USUBJID
            TEST <- SENDstudy$lb[which(SENDstudy$lb$USUBJID %in% Subjs), c("LBDY","LBSTRESN","LBTESTCD","USUBJID")]
            Test3 <- TEST[which(TEST$LBTESTCD %in% c("SPGRAV","VOLUME")),]
            for (day in unique(Test3$LBDY)){
                Test4 <- Test3[which(Test3$LBDY == day), c("USUBJID","LBTESTCD","LBSTRESN")]
                Test4 <- reshape(Test4, idvar = "USUBJID", timevar = "LBTESTCD", direction = "wide")
                #Model of relationship
                newx <- seq(min(Test4$LBSTRESN.SPGRAV), max(Test4$LBSTRESN.SPGRAV), by = 0.001)
                ydata <- seq(min(Test4$LBSTRESN.VOLUME), max(Test4$LBSTRESN.VOLUME),
                             by = (max(Test4$LBSTRESN.VOLUME)-min(Test4$LBSTRESN.VOLUME))/length(newx))
                ydata <- sort(ydata, decreasing = TRUE)
                if (length(ydata)>length(newx)){
                    ydata <- ydata[1:length(newx)]
                }
                #Model attempt 1 - linear
                # modelline <- lm(ydata ~ exp(-1/newx))
                # newxy <- list(newx)
                # modeltest <- predict(modelline)
                # modeldata <- data.frame(y = modeltest, x = newx)
                #model attempt 2 - nonlinear
                fo3 <- y ~ 1/(x^c)
                data <- data.frame(y = ydata, x = newx)
                fm3 <- nls(fo3, data=data, start = list(c=1))
                modeltest <- predict(fm3)
                modeldata <- data.frame(y = sort(modeltest,decreasing = TRUE), x = newx)

                #add in gender
                Test4 <- merge(Test4, ExampleSubjects[,c("USUBJID","SEX","ARM")], by = "USUBJID")
                q <- ggplot() +
                    geom_point(data = Test4, aes(x=LBSTRESN.SPGRAV, y=LBSTRESN.VOLUME, shape = SEX, color = ARM)) +
                    geom_line(data = modeldata, aes(x=x, y=y, color = 'Model of Relationship'))+
                    labs(x='Urine Specific Gravity (SPGRAV)', y="Urine Volume (mL)") +
                    ggtitle(paste0("Generated Volume and SPGRAV for day ", day, " of collection"))
                print(q)
            }


            #Coordinate LBORRES and LBSTRESC
            SENDstudy$lb$LBSTRESC <- as.character(SENDstudy$lb$LBSTRESN)
            SENDstudy$lb$LBORRESU <- SENDstudy$lb$LBSTRESU
            SENDstudy$lb$LBORRES <- SENDstudy$lb$LBSTRESN

            #Remove LBBLFL
            SENDstudy$lb$LBBLFL <- NA

            #Make Factors as Characters
            SENDstudy$lb$LBTESTCD <- as.character(SENDstudy$lb$LBTESTCD)
            SENDstudy$lb$LBTEST <- as.character(SENDstudy$lb$LBTEST)
            if (any(grepl('LBCAT',colnames(SENDstudy$lb)) == TRUE)){
                SENDstudy$lb$LBCAT <- as.character(SENDstudy$lb$LBCAT)
            }
            if (any(grepl('LBSCAT',colnames(SENDstudy$lb)) == TRUE)){
                  SENDstudy$lb$LBSCAT <- as.character(SENDstudy$lb$LBSCAT)
            }
            SENDstudy$lb$LBMETHOD <- as.character(SENDstudy$lb$LBMETHOD)
            SENDstudy$lb$LBSPEC <- as.character(SENDstudy$lb$LBSPEC)
            SENDstudy$lb$LBSTRESU <- as.character(SENDstudy$lb$LBSTRESU)

            #Generate MI Data
            #Keeps: MISPEC, MIDY, MISTESTCD, MITEST,
            #Replaces: STUDYID, USUBJID, MIDTC, MISTRESC, MIORRES, MISEV
            #Removes: MIREASND, MISPCCND,MISPCUFL, MIDTHREL and MIREASND


            #Replace StudyID and USUBJID
            SENDstudy$mi$STUDYID <- rep(studyID, nrow(SENDstudy$mi))
            SENDstudy$mi$USUBJID <- as.character(SENDstudy$mi$USUBJID)
            SENDstudy$mi$MIORRES <- as.character(SENDstudy$mi$MIORRES)
            SENDstudy$mi <- merge( USUBJIDTable,SENDstudy$mi, by = "USUBJID")
            SENDstudy$mi <- SENDstudy$mi[,!(names(SENDstudy$mi) %in% "USUBJID")]
            names(SENDstudy$mi)[names(SENDstudy$mi) == "NEWUSUBJID"] <- "USUBJID"
            #Remove Dates
            cols <- grep("DTC", colnames(SENDstudy$mi))
            SENDstudy$mi[,cols] <- rep("XXXX-XX-XX",length(SENDstudy$mi$STUDYID))
            #Consolidate "Normal" Findings
            Example$mi$MISTRESC <- as.character(Example$mi$MISTRESC)
            SENDstudy$mi$MISTRESC <- as.character(SENDstudy$mi$MISTRESC)
            Example$mi$MISTRESC <- toupper(Example$mi$MISTRESC)
            Example$mi$MISTRESC <-  str_replace_all(Example$mi$MISTRESC, "NORMAL", "UNREMARKABLE")
            Example$mi$MISTRESC <-  str_replace_all(Example$mi$MISTRESC, "NAD", "UNREMARKABLE")
            Example$mi$MISTRESC <-  str_replace_all(Example$mi$MISTRESC, "NO ABNORMALITY DETECTED", "UNREMARKABLE")
            Example$mi$MISTRESC <-  str_replace_all(Example$mi$MISTRESC,"NO ABUNREMARKABLEITY DETECTED","UNREMARKABLE")
            #Find out investigated MISPECS and Frequency of Findings/Severity Range of Findings
            #Calculate percentage of Findings per MISPEC in each ARM
            MIFindings <- merge(Subjects, Example$mi[,c("USUBJID", "MISPEC", "MISTRESC","MISEV")], by = "USUBJID")
            FindingsPercen <- MIFindings %>%
                group_by(Dose, MISPEC,SEX) %>%
                count(MISTRESC) %>%
                dplyr::mutate(percent = n/sum(n)) %>%
                dplyr::select(-n)
            SevPercen <- MIFindings %>%
                group_by(Dose, MISPEC,SEX) %>%
                count(MISEV) %>%
                dplyr::mutate(percent = n/sum(n)) %>%
                dplyr::select(-n)
            SENDstudy$mi$MIDY <- as.numeric(SENDstudy$mi$MIDY)

            #How to incorporate MCMC in mi....
                # Correspond across which organs... SEverity could be done by regression, but not MISPEC


            #Simulate MISPEC Results to create representative distributions of values
            for(Dose in unique(Doses$Dose)){
                for (gender in unique(ExampleSubjects$SEX)){
                    Subjs <- ExampleSubjects$USUBJID[which(ExampleSubjects$ARM == Dose  & ExampleSubjects$SEX == gender)]
                    GroupTests <- SENDstudy$mi$MISPEC[which(SENDstudy$mi$USUBJID %in% Subjs)]
                    for (Organ in unique(GroupTests)){
                        SevPerc <- SevPercen[which(SevPercen$Dose == Dose & SevPercen$MISPEC == Organ), c('MISEV','percent')]
                        FindPercen <- FindingsPercen[which(FindingsPercen$Dose == Dose & FindingsPercen$MISPEC == Organ), c('MISTRESC','percent')]
                        idx <- which(SENDstudy$mi$USUBJID %in% Subjs & SENDstudy$mi$MISPEC %in% Organ )
                        #Create Appropriate Findings Per MISPEC
                        FindGendata <- sample(FindPercen$MISTRESC,length(idx), replace = TRUE, prob = FindPercen$percent)
                        #Create Appropriate Severity per MISPEC
                        GenData <- data.frame('MISTRESC' = FindGendata,
                                              'MISEV'  = rep(NA, length(FindGendata)),
                                              'MIORRES' = rep(NA, length(FindGendata)))
                        lvls <- levels(Example$mi$MISEV)
                        GenData$MISEV <- factor(GenData$MISEV, levels = lvls)
                        for (i in 1:nrow(GenData)){
                            GenData$MIORRES[i] <- paste0(Organ, ": ", GenData$MISTRESC[i])
                            if (GenData$MISTRESC[i] %in% c("UNREMARKABLE",NA)){
                                GenData$MISEV[i] <- NA
                            } else {
                                Sevs <- as.data.frame(SevPerc)[which(is.na(SevPerc$MISEV) == FALSE),]
                                if (length(Sevs$MISEV)==0){
                                    #Check if all of the Findings have NA Severity
                                    GenData$MISEV[i] <- NA
                                }else{
                                    GenData$MISEV[i] <- sample(Sevs$MISEV,1,replace = TRUE, prob = Sevs$percent)
                                }
                            }
                        }
                        #Store Generated Values
                        SENDstudy$mi[idx,c("MISTRESC","MISEV","MIORRES")] <-GenData
                    }
                    #Make MIDY Appropriate
                    MIDY <- max(SENDstudy$mi[which(SENDstudy$mi$USUBJID %in% Subjs), "MIDY"], na.rm = TRUE)
                    SENDstudy$mi[which(SENDstudy$mi$USUBJID %in% Subjs), "MIDY"] <- MIDY
                }
            }
            #Make MIRESCAT for Non-Normal Findings
            SENDstudy$mi$MIRESCAT <- NA
            SENDstudy$mi$MIRESCAT[which(grepl("UNREMARKABLE",SENDstudy$mi$MISTRESC) == FALSE)] <- "NON-NEOPLASTIC"

            #Remove MISTAT, MIRESCAT, MIDTHREL, and MISPCCND
            SENDstudy$mi$MISTAT <- NA
            SENDstudy$mi$MIREASND <- NA
            SENDstudy$mi$MISPCCND <- NA
            SENDstudy$mi$MISPCUFL <- NA
            SENDstudy$mi$MIDTHREL <- NA

            #Make Factors as Characters
            SENDstudy$mi$MITESTCD <- as.character(SENDstudy$mi$MITESTCD)
            SENDstudy$mi$MITEST <- as.character(SENDstudy$mi$MITEST)
            SENDstudy$mi$MISPEC <- as.character(SENDstudy$mi$MISPEC)
            SENDstudy$mi$MISTRESC <- as.character(SENDstudy$mi$MISTRESC)
            SENDstudy$mi$MISEV <- as.character(SENDstudy$mi$MISEV)

            ##### Save Generated Study for Tables ####
            GeneratedSEND[[j]] <- SENDstudy

            ############# Export Folder of Generated Data #################

            if (PRINT == TRUE){
                #Create .xpt files if you can
                dir.create(paste0('FAKE',studyID)) #Create Folder to Hold Study

                #Loop through domains created to print them in created folder
                Domains <- names(SENDstudy)
                for (domain in Domains){
                    printpath <- paste0(homePath,"/FAKE",studyID,"/",domain,".xpt")
                    write_xpt(SENDstudy[[domain]],path = printpath, version = 5)
                }
            }


        }

        #Create Fused DataFrame from GeneratedSEND
        GeneratedALL <- list()
        GeneratedALL$bw <- as.data.frame(GeneratedSEND[[1]]$bw)
        GeneratedALL$dm <- as.data.frame(GeneratedSEND[[1]]$dm)
        GeneratedALL$ds <- as.data.frame(GeneratedSEND[[1]]$ds)
        GeneratedALL$ex <- as.data.frame(GeneratedSEND[[1]]$ex)
        GeneratedALL$lb <- as.data.frame(GeneratedSEND[[1]]$lb)
        GeneratedALL$mi <- as.data.frame(GeneratedSEND[[1]]$mi)
        GeneratedALL$ta <- as.data.frame(GeneratedSEND[[1]]$ta)
        GeneratedALL$ts <- as.data.frame(GeneratedSEND[[1]]$ts)
        GeneratedALL$tx <- as.data.frame(GeneratedSEND[[1]]$tx)
        if (Createme >1){
            for (t in 2:Createme){
                GeneratedALL$bw <- rbind(GeneratedALL$bw, GeneratedSEND[[t]]$bw)
                GeneratedALL$dm <- rbind(GeneratedALL$dm, GeneratedSEND[[t]]$dm)
                GeneratedALL$ds <- rbind(GeneratedALL$ds, GeneratedSEND[[t]]$ds)
                GeneratedALL$ex <- rbind(GeneratedALL$ex, GeneratedSEND[[t]]$ex)
                GeneratedALL$lb <- rbind(GeneratedALL$lb, GeneratedSEND[[t]]$lb)
                GeneratedALL$mi <- rbind(GeneratedALL$mi, GeneratedSEND[[t]]$mi)
                GeneratedALL$ta <- rbind(GeneratedALL$ta, GeneratedSEND[[t]]$ta)
                GeneratedALL$ts <- rbind(GeneratedALL$ts, GeneratedSEND[[t]]$ts)
                GeneratedALL$tx <- rbind(GeneratedALL$tx, GeneratedSEND[[t]]$tx)
            }
        }

        #Display generated .xpt files as datatables
        output$BW <- renderDataTable(GeneratedALL$bw, options = list(pageLength = 15))
        output$DM <- renderDataTable(GeneratedALL$dm, options = list(pageLength = 15))
        output$DS <- renderDataTable(GeneratedALL$ds, options = list(pageLength = 15))
        output$EX <- renderDataTable(GeneratedALL$ex, options = list(pageLength = 15))
        output$LB <- renderDataTable(GeneratedALL$lb, options = list(pageLength = 15))
        output$MI <- renderDataTable(GeneratedALL$mi, options = list(pageLength = 15))
        output$TA <- renderDataTable(GeneratedALL$ta, options = list(pageLength = 15))
        output$TS <- renderDataTable(GeneratedALL$ts, options = list(pageLength = 15))
        output$TX <- renderDataTable(GeneratedALL$tx, options = list(pageLength = 15))

    })


}

# Run the application
shinyApp(ui = ui, server = server)