1_get_raw_data.yml

target_default: 1_get_raw_data

packages:
  - dplyr
  - readxl
  - tidyr
  
sources:
  - 1_get_raw_data/src/get_raw_data.R
  
targets:
  1_get_raw_data:
    depends:
      - raw_sample
      - raw_site
      - raw_5507
      - raw_5433
      - pct_rec_mspikes
      - pct_rec_labcontrol
      - pct_rec_mspikes_5507
      - pct_rec_mspikes_5433
      - procedural_blanks
      - lab_duplicates
  
  raw_sample: # read in raw sample concentration data from Batelle
    command: get_samples(file = '1_get_raw_data/raw_data/GLEI CEC Study Data_05Oct2017.xlsx')
    
  raw_site: # read in site metadata for all sites in GLRI study
    command: get_sites(file = '1_get_raw_data/raw_data/GLEI CEC Study Sites.csv')
    
  raw_5507: # get MKE study data using dataRetrieval, uses pah::get_pah
    command: get_mke_data()
    
  raw_5433:
    command: get_glri6_data(all_sites = raw_site)
    
  procedural_blanks:
    command: get_qa_data(qa_type = I('pblanks'))
    
  lab_duplicates:
    command: get_qa_data(qa_type = I('labdups'))
    
  pct_rec_mspikes:
    command: get_qa_data(qa_type = I('mspikes'))
    
  pct_rec_labcontrol:
    command: get_qa_data(qa_type = I('labcontrol'))
    
  pct_rec_mspikes_5507:
    command: read.csv(file = '1_get_raw_data/raw_data/NWQL_prepSpike_recoveries.csv', 
                           skip = I(2), stringsAsFactors = FALSE)
                           
  pct_rec_mspikes_5433:
    command: read.csv(file = '1_get_raw_data/raw_data/5433 Set Spike recoveries 0101_2017 to 0101_2018 prep dates v042418.csv', 
                           skip = I(2), stringsAsFactors = FALSE)