inst/dataprep/BCG-CD4.R

In MiguelRodo/VaccCompData: Data package for novel TBV vaccine comparison paper

# Loading -----------------------------------------------------------------

# names of current objects
currObjVec = ls()

# find library containing raw data
pkgPath = system.file( package = "VaccCompData" )

# libraries
source( file = paste0( pkgPath, "/dataprep/libraries.R") )

#read in data
fileLoc = paste0( pkgPath, "/extdata/bcg.xlsx" )
cd4Tbl = as_tibble( read.xlsx( fileLoc, sheet = 'CD4') )

# Data Prep I ---------------------------------------------------------------

### Goal: Put data into a format appropriate for later manipulation.
### That means that only multiple cytokine combinations are allowed,
### values under variables are consistent ( e.g. always OBI and never OBi ),
### and there are no unnecessary columns.

###################
### DATA BASICS ###
###################

nCol = ncol( cd4Tbl )

##############################
### NO UNNECESSARY COLUMNS ###
##############################

# cd4Tbl %<>% select( -c( Subset, X49:X52 ) )
cd4Tbl %<>% select( -Subset )
# exclude Subset and X49:X52
# X49 to X52 not found in raw data, and Subset not useful since
# CD4 or CD8 is specified later and differently.

#########################
### CONSISTENT VALUES ###
#########################

#unique values in id info
unique(cd4Tbl$Stimulation) #"BCG"   "ECppl" "PHA"   "UNS". Correct
unique(cd4Tbl$Study.Day) #"SC-3" "V01"  "V08"  "V13"  "V15"  "V28"  "V20"  "V23"  "V30". Formatting seems fine.
unique(cd4Tbl$Group) # "IBO" "iBO" "OBI" "oBI". Incorrect
cd4Tbl$Group = str_to_upper(cd4Tbl$Group) #correct above
unique(cd4Tbl$Group) # "IBO" "OBI". Corrected

# Sample
cd4Tbl$Sample = str_to_upper( cd4Tbl$Sample ) #all caps

#Stimulation
cd4Tbl$Stimulation = ifelse( str_detect( cd4Tbl$Stimulation, 'ECppl' ) == TRUE, 'ECPP',  cd4Tbl$Stimulation)

# Remove Non-AN
cd4Tbl %<>%
  mutate( rowNumber = 1:n() ) %>%
  group_by( rowNumber ) %>%
  do( removeNonANEntries( ., columnNameVec = c( "Sample", "Study.Day" ) ) ) %>%
  ungroup() %>%
  select( - rowNumber)

######################
### ONLY MULTIPLES ###
######################

cd4Tbl %<>%
  gather( key = cytCombo, value = frequency, `CD4+/2+`:`CD4+/2-17-22-g-T-` ) %>%
  mutate( cytCombo = str_to_upper( cytCombo ) ) %>%
  filter( str_length( cytCombo) > 10  ) %>%
  spread( key = cytCombo, value = frequency ) %>%
  arrange( Group, PTID, Study.Day, Stimulation )

#######################
### RENAME BOOLEANS ###
#######################

### Goal: Put dataTibble into format where cytokine combinations
### are consistently labelled with a p/n format, in a specific order.

### CHECK LENGTHS OF VARIABLE NAMES
unique ( str_length( tbl_vars( cd4Tbl )[ 6: ncol( cd4Tbl ) ] ) ) #all 17

### CHANGE POSITIVITY INDICATOR

cd4Tbl = replaceCytPosIndicator( dataTibble = cd4Tbl, firstMultCytIndex = 6,
                                      posPattern = '+', negPattern = '-',
                                      posRep = "p", negRep = "n",
                                      alphaNumericOrigSign = TRUE,
                                      signLocVec = c( 7, 10, 13, 15, 17 ) )

### RE-ORDER

cd4Tbl = changeCytOrder( dataTibble = cd4Tbl, firstCytColIndex = 6,
                          firstCytNameLoc = 6, signLocVec = c( 7, 10, 13, 15, 17 ),
                          orderLocVec = c( 2, 4, 5, 1, 3 ), cdPresent = TRUE )

# Data Checking -----------------------------------------------------------

#Goal: Check for any errors that can be detected detect, and correct these.

####################
### STIMULATIONS ###
####################

# Goal: Check that the stimulation listed in the Sample column is
# the same as the stimulation listed in the Stimulation column,
# and there is a BCG and UNS for each study day and ptid combo.

### MATCHING STIMULATION AND SAMPLE

nonMatchingStimRowsList = compToFullName( dataTibble = cd4Tbl, compColumnName = 'Stimulation', mainColumnName = 'Sample' ) # BCG - 1510. PHA - 1510.

cd4Tbl[ nonMatchingStimRowsList$BCG, ] # "BCG"  "PHA"  "170176OBIV20BCGFCS". BCG in Sample, but PHA in Stim
#print(cd4Tbl %>% filter(PTID == 'ID_0176' & Study.Day == "V20" ), n = 28) # two PHAs in same day
cd4Tbl$Stimulation[ nonMatchingStimRowsList$BCG ] = 'BCG'
compToFullName( dataTibble = cd4Tbl, compColumnName = 'Stimulation ', mainColumnName = 'Sample' ) # All NULL.

##################
### STUDY DAYS ###
##################

### MATCHING STIMULATION AND SAMPLE

nonMatchingStudyDaysRowsList = compToFullName( dataTibble = cd4Tbl, compColumnName = 'Study.Day', mainColumnName = 'Sample' ) # BCG - 1510. PHA - 1510.

cd4Tbl[ nonMatchingStudyDaysRowsList$V13, ] # V13 in Study.Day, but V15 in Sample
#print(cd4Tbl %>% filter(PTID == 'ID_0063' & Study.Day == "V13" ), n = 28) # two V13s for same person in Study.Day
cd4Tbl$Study.Day[ nonMatchingStudyDaysRowsList$V15 ] = 'V15'
compToFullName( dataTibble = cd4Tbl, compColumnName = 'Stimulation ', mainColumnName = 'Sample' ) # All NULL.

### STIMULATIONS BY STUDY DAY AND PTID COMBO

stimCountDF = cd4Tbl %>%
  group_by(PTID, Study.Day) %>%
  summarise(stimCount = n(), bcgCount = sum(str_detect(Stimulation, 'BCG') * 1), unsCount = sum(str_detect(Stimulation, 'UNS') * 1))
unique(stimCountDF$stimCount) #4. Each PTID and study day has four stims each.
unique(stimCountDF$bcgCount) #1. Each PTID and Study day has one BCG
unique(stimCountDF$unsCount) #1. Each PTID and study day has one UNS

# Calculation -------------------------------------------------------------

#####################
### SUM OVER IL22 ###
#####################

cd4Tbl = cd4Tbl %>%
  gather( key = cytCombo, value = resp, CD4Gn2nTn17n22n:CD4Gp2pTp17p22p ) %>%
  group_by( PTID, Stimulation, Study.Day, Group ) %>%
  mutate( cytCombo = str_sub( cytCombo, end = -4  )) %>%
  group_by( Sample, PTID, Stimulation, Study.Day, Group, cytCombo ) %>%
  summarise( resp = sum( resp ) ) %>%
  ungroup() %>%
  spread( cytCombo, resp )

########################
### SUB UNS FROM BCG ###
########################

# test that the order of stims is as expected
testStr = cd4Tbl %>%
  group_by( PTID, Study.Day ) %>%
  arrange( PTID, Study.Day, Stimulation ) %>%
  summarise( stim = paste0( Stimulation[1], Stimulation[2],
    Stimulation[3], Stimulation[4] ) ) %>%
  ungroup() %>%
  `[[`( "stim" ) %>%
  unique()

if( !identical( testStr, "BCGECPPPHAUNS" ) ) stop( "cd8Tbl does not have the same stims, or the same order of stims, for each individual." )

cd4Tbl %<>%
  group_by( PTID, Study.Day ) %>%
  arrange( PTID, Study.Day, Stimulation ) %>%
  mutate_if( is.numeric, function(x) x[1] - x[4] ) %>%
  slice(1) %>%
  ungroup()

cd4ITbl = cd4Tbl %>%
  rename( stim = Stimulation ) %>%
  select( -Sample )

cd4Tbl %<>%
  select( -c( Sample, Stimulation ) ) %>%
  ungroup()

# Data Prep II -------------------------------------------------------------

#####################
### COLUMN RENAME ###
#####################

cd4ITbl %<>%
  rename( ptid = PTID, group = Group )

cd4Tbl %<>%
  rename( ptid = PTID, group = Group )

##################################
### CONVERT STUDY DAYS TO DAYS ###
##################################

#l4ITbl = cd4ITbl %>%
#  gather( key = cytCombo, value = frequency, CD4Gn2nTn17n:CD4Gp2pTp17p ) %>%
#  mutate( timePoint = convTimePoint( Study.Day ) ) %>%
#  select( - Study.Day ) %>%
#  filter( timePoint != - 1)  %>%
#  filter( ptid != 'ID_0102' & ptid != 'ID_0104' & ptid != 'ID_0131' & ptid != 'ID_0250' )

l4ITbl = cd4ITbl %>%
  gather( key = cytCombo, value = frequency, CD4Gn2nTn17n:CD4Gp2pTp17p ) %>%
  mutate( timePoint = convTimePoint( Study.Day ) ) %>%
  select( - Study.Day ) %>%
  filter( timePoint != - 1)  %>%
  filter( ptid != 'bcg_31' & ptid != 'bcg_12' & ptid != 'bcg_17' & ptid != 'bcg_30' )

l4Tbl = cd4Tbl %>%
  gather( key = cytCombo, value = frequency, CD4Gn2nTn17n:CD4Gp2pTp17p ) %>%
  mutate( timePoint = convTimePoint( Study.Day ) ) %>%
  select( - Study.Day ) %>%
  filter( timePoint != - 1)  %>%
  filter( ptid != 'bcg_31' & ptid != 'bcg_12' & ptid != 'bcg_17' & ptid != 'bcg_30' )

# make stim lower case
l4ITbl %<>% mutate( stim = 'bcg' )


# Testing -----------------------------------------------------------------

# l4ITbl
testVal1 = l4ITbl %>%
  filter( ptid == 'bcg_8' & timePoint == 365 & stim == "bcg" & cytCombo == "CD4Gn2pTn17n" ) %>%
  extract2("frequency")
if( testVal1 != ( 0.003385 - 0 + 0.052802 - 0.007228 ) ) stop( "l4ITbl error")
testVal2 = l4ITbl %>%
  filter( ptid == 'bcg_1' & timePoint == 0 & stim == "bcg" & cytCombo == "CD4Gn2nTp17n" ) %>%
  extract2("frequency")
if( testVal2 != ( 0.000614 + 0.155341 - 0.016429 ) ) stop( "l4ITbl error" )

# l4Tbl
testVal1 = l4Tbl %>%
  filter( ptid == 'bcg_8' & timePoint == 365 & cytCombo == "CD4Gn2pTn17n" ) %>%
  extract2("frequency")
if( testVal1 != ( 0.003385 - 0 + 0.052802 - 0.007228 ) ) stop( "l4Tbl error")
testVal2 = l4Tbl %>%
  filter( ptid == 'bcg_1' & timePoint == 0 & cytCombo == "CD4Gn2nTp17n" ) %>%
  extract2("frequency")
if( testVal2 != ( 0.000614 + 0.155341 - 0.016429 ) ) stop( "l4Tbl error" )

# Clear Workspace ---------------------------------------------------------

# remove all objects but initial objects (currObjVec) and processed data tibble (l4Tbl)
rm( list = setdiff( ls(), c( currObjVec, "l4Tbl", "l4ITbl" ) ) )