tremorModule.R

# Entry function to extract tremor features

######################
# Load Libraries
######################
library(optparse)
library(data.table)
library(plyr)
library(tidyverse)

library(jsonlite)
library(uuid)
library(lubridate)

library(synapser)
library(mpowertools) 
# devtools::install_github("Sage-Bionetworks/mpowertools")

synLogin(Sys.getenv("synapseUsername"), 
         Sys.getenv("synapsePassword"))
batch_job_number <- Sys.getenv("AWS_BATCH_JOB_ARRAY_INDEX")


###########################
# Synapse I/O parameters
###########################
readArgs <- function() {
  parser <- OptionParser(description='Extract tremor features.')
  parser <- add_option(parser, "--jobTable", type="character",
                       help=paste("Synapse Table containing the record IDs",
                                  "to be processed."))
  parser <- add_option(parser, "--inputTable", type="character",
                       help=paste("Synapse Table containing columns with raw",
                                  "features to extract."))
  parser <- add_option(parser, "--assayName", type="character",
                       help="Name of assay being processed")
  parser <- add_option(parser, "--assayColumn", type="character",
                       help="Name of column with assay json files")
  parser <- add_option(parser, "--outputPath", type="character",
                       help="Directory to write results to.")
  parse_args(parser)
}

# Set input parameters
args <- readArgs()
jobTable <- synTableQuery(paste0("select jobInfo from ", args$jobTable,
                        " where uid = ", batch_job_number))$asDataFrame()
rid.to.extract <- fromJSON(jobTable$jobInfo)$recordIds

# Get source tremor table from synapse
rid.to.extract.str.list <- paste0("('",
                                  paste(rid.to.extract, collapse="', '"),
                                  "')")
query <- paste("select * from", args$inputTable,
               "where recordId in", rid.to.extract.str.list)
print(paste("Querying", query))
tremor.table <- synTableQuery(query)
                                    
tremor.table.values <- tremor.table$asDataFrame()


########################
# Feature extraction
########################
# Download tremor json files
print(paste("Downloading json files for", args$assayName))
if (dim(tremor.table.values)[1] > 0) {
  get.tremor.json.loc <- function(col.name, tremorTable) {
      tremor.table.values[col.name] <- as.character(
                                        tremor.table.values[,col.name])
      tbl.files <- synDownloadTableColumns(tremorTable, col.name) %>%
                   lapply(function(x) data.frame(V1=x)) %>% 
                   rbindlist(idcol=col.name) %>% 
                   plyr::rename(c('V1'=gsub('.items',
                                            '.fileLocation',
                                            col.name))) %>%
               dplyr::left_join(tremor.table.values, by=col.name)
      return(tbl.files)
      }
  tremor.json.loc <- get.tremor.json.loc(args$assayColumn, tremor.table)
} else {
  tremor.json.loc <- data.frame()
}

## Set input parameters
WORKING_DIR = '/mPower_efs/mPowerAnalysis/featureExtraction/'
LIBRARY_DIRECTORY = '/work/mylibs/'
CACHE_DIR = '/mPower_efs/.synapseCache/'
TREMOR_TBL = 'syn10676309' # mPower public tremor table from Sage-internal project 
TREMOR_FILE_LOC_ID = 'syn10916227' # files containing the json fle location in /mPower_efs 
FEATURES_PARENT_ID = 'syn8362851' # Destination directory in synapse

UPDATE = FALSE
n = 1000

PARAM = list(
  handInLapLeft = list(FEATURES_ID = 'syn11597056', # Destination file id where features are to be stored
                       COLUMN_NAME = 'deviceMotion_tremor_handInLap_left.json.items', # Source column name of synapse table to obtain json files
                       COLUMN_NAME1 = 'deviceMotion_tremor_handInLap_right.json.items'), # Comparison column name of synapse table to obtain json files
  handInLapRight = list(FEATURES_ID = 'syn11597058', # Destination file id where features are to be stored
                        COLUMN_NAME = 'deviceMotion_tremor_handInLap_right.json.items',
                        COLUMN_NAME1 = 'deviceMotion_tremor_handInLap_left.json.items'),
  handAtShoulderLengthLeft = list(FEATURES_ID = 'syn11597059', # Destination file id where features are to be stored,
                                  COLUMN_NAME = 'deviceMotion_tremor_handAtShoulderLength_left.json.items',
                                  COLUMN_NAME1 = 'deviceMotion_tremor_handAtShoulderLength_right.json.items'),
  handAtShoulderLengthRight = list(FEATURES_ID = 'syn11597060', # Destination file id where features are to be stored
                                   COLUMN_NAME = 'deviceMotion_tremor_handAtShoulderLength_right.json.items',
                                   COLUMN_NAME1 = 'deviceMotion_tremor_handAtShoulderLength_left.json.items'),
  handToNoseLeft = list(FEATURES_ID = 'syn11597061', # Destination file id where features are to be stored
                        FNAME = 'handToNoseLeftTremorFeatures.tsv',
                        COLUMN_NAME = 'deviceMotion_tremor_handToNose_left.json.items'),
  handToNoseRight = list(FEATURES_ID = 'syn11597062', # Destination file id where features are to be stored
                         FNAME = 'handToNoseRightTremorFeatures.tsv',
                         COLUMN_NAME = 'deviceMotion_tremor_handToNose_right.json.items')
)

#################
# Github commit
#################
thisFileName = 'tremorModule.R'

thisRepo <- getRepo(repository = "th1vairam/mPowerAnalysis", ref="branch", refName='tremor')
thisFile <- getPermlink(repository = thisRepo, repositoryPath=paste0('featureExtraction/',thisFileName))
thisRepo <- getRepo(repository = "th1vairam/mpowertools", ref="branch", refName='tremor_dev')
thisFile1 <- getPermlink(repository = thisRepo, repositoryPath=paste0('R/','getTremorFeatures.R'))
thisFile2 <- getPermlink(repository = thisRepo, repositoryPath=paste0('R/','getKineticTremorFeatures.R'))
thisFile3 <- getPermlink(repository = thisRepo, repositoryPath=paste0('R/','getComparativeTremorFeatures.R'))

##############################

# Extract features
print(paste("Extracting features for", args$assayName))
if(dim(tremor.json.loc)[1] != 0) {
  assayList <- c('handInLapLeft','handInLapRight',
                 'handAtShoulderLengthLeft', 'handAtShoulderLengthRight')
  if (any(args$assayName %in% assayList)) {
    ftr <- tremor.json.loc %>%
           plyr::dlply(.(recordId),
                       .fun=function(x, columnName) {
                         mpowertools::getTremorFeatures(
                           x[1, gsub('.items',
                                     '.fileLocation',
                                     columnName)])},
                       args$assayColumn) %>%
           data.table::rbindlist(idcol='recordId', use.names=T, fill=T) %>%
           dplyr::mutate(Assay=args$assayName)
  } 
  else if (any(args$assayName %in% c('handToNoseLeft','handToNoseRight'))) {
      ftr <- tremor.json.loc %>%
             plyr::dlply(.(recordId),
                         .fun=function(x, columnName) {
                           mpowertools::getKineticTremorFeatures(
                             x[1,gsub('.items',
                                      '.fileLocation',
                                      columnName)])},
                         args$assayColumn) %>%
             data.table::rbindlist(idcol='recordId', use.names=T, fill=T) %>%
             dplyr::mutate(Assay=args$assayName)
  }
  
  # Download tremor json files
  if (dim(tremor.table@values)[1] > 0){
    if (any(assay %in% c('handInLapLeft','handInLapRight', 'handAtShoulderLengthLeft', 'handAtShoulderLengthRight'))){
      tremor.json.loc = lapply(para[c('COLUMN_NAME', 'COLUMN_NAME1')], function(col.name, tremorTable){
        tbl.files = synDownloadTableColumns(tremorTable, col.name) %>%
          lapply(function(x) data.frame(V1 = x)) %>% 
          rbindlist(idcol = col.name) %>% 
          plyr::rename(c('V1' = gsub('.items','.fileLocation', col.name))) %>%
          dplyr::left_join(tremorTable@values, by = col.name)
        }, tremor.table) %>%
        plyr::join_all(type = 'full', by = c('recordId', 'healthCode'))
    } else {
      tremor.json.loc = lapply(para[c('COLUMN_NAME')], function(col.name, tremorTable){
        tbl.files = synDownloadTableColumns(tremorTable, col.name) %>%
          lapply(function(x) data.frame(V1 = x)) %>% 
          rbindlist(idcol = col.name) %>% 
          plyr::rename(c('V1' = gsub('.items','.fileLocation', col.name))) %>%
          dplyr::left_join(tremorTable@values, by = col.name)
      }, tremor.table) %>%
        plyr::join_all(type = 'full', by = c('recordId', 'healthCode'))
    }
  } else {
    tremor.json.loc = data.frame()

  }
  parent.directory = paste0(args$outputPath, "/", args$assayName)
  if (!dir.exists(parent.directory)) {
      dir.create(parent.directory)
  }
  fpath=paste0(parent.directory, "/", uuid::UUIDgenerate(), '.tsv')
  data.table::fwrite(ftr, file=fpath, append=FALSE,
                     row.names=FALSE, quote=FALSE, sep='\t')
  }
print(paste('Finished', args$assayName, batch_job_number))