• Calculate Compound-Level Polyphenol Intakes
• SCRIPTS
  • Specify grouping variables
  • Daily Class Polyphenol Intake Numbers BY ENTRY (Record/Recall)
  • Daily Class Intakes by Subject

Calculate Compound-Level Polyphenol Intakes

This script calculates compound-level polyphenol intake (mg, mg/1000kcal) for provided dietary data.

INPUTS

• Diet_FooDB_polyphenol_content.csv.bz2: Disaggregated dietary data, mapped to FooDB polyphenol content, at the compound-level
• Diet_total_nutrients.csv - total daily nutrient data to go with dietary data.
• Diet_polyphenol_classtax_3072.csv - class taxonomy is derived from FooDB which uses ClassyFire, an automated chemical taxonomic classification application based on chemical structure

OUTPUTS

• summary_compound_intake_by_entry.csv, polyphenol compound intakes by recall for each participant
• summary_compound_intake_by_subject.csv, polyphenol compound intakes for each participant, provided in long format (compounds as rows)
• summary_compound_intake_by_subject_wide.csv, polyphenol compound intakes for each participant, provided in wide format (compounds as columns)

SCRIPTS

# Load packages
suppressMessages(library(dplyr))
suppressMessages(library(vroom))
suppressMessages(library(tidyr))
suppressMessages(library(stringr))

# Load provided file paths
source("provided_files.R")

#Content and kcal data
input_polyphenol_content = vroom::vroom('outputs/Diet_FooDB_polyphenol_content.csv.bz2',
                                        show_col_types = FALSE)

input_kcal = vroom::vroom('outputs/Diet_total_nutrients.csv', show_col_types = FALSE) %>%
  # Ensure consistent KCAL naming whether ASA24 or NHANES
  dplyr::rename_with(~ "Total_KCAL", .cols = any_of(c("Total_KCAL", # Specific to ASA24
                                               "Total_DRXIKCAL"))) %>%  # Specific to NHANES
  dplyr::select(c(subject, 
           # Ensures we pull correct columns for record or recall
           any_of(c("RecallNo", "RecordNo", "RecordDayNo")),
           Total_KCAL))

# Class taxonomy for FooDB compounds
class_tax = vroom::vroom(class_tax, show_col_types = FALSE) %>%
  dplyr::select(c(compound_public_id, class))

# Merge the files
input_polyphenol_kcal = dplyr::left_join(input_polyphenol_content, input_kcal) %>%
  # Add class data
  left_join(class_tax)

## Joining with `by = join_by(subject, RecallNo)`
## Joining with `by = join_by(compound_public_id)`

Specify grouping variables

Column grouping depends on whether output is from a record or recall.

if ("RecallNo" %in% names(input_polyphenol_kcal)) {
  group_vars = c("subject", "RecallNo", "compound_public_id")
  
} else if ("RecordNo" %in% names(input_polyphenol_kcal)) {
  group_vars = c("subject", "RecordNo", "RecordDayNo", "compound_public_id")
  
} else {
  stop("Data must contain RecallNo or RecordNo.")
}

Daily Class Polyphenol Intake Numbers BY ENTRY (Record/Recall)

compound_intakes_entry = input_polyphenol_kcal %>%
  
  # Recall - Sum by Subject, Recall
  # Record - Sum by Subject, Record Number, Day in Record Number
  # Both recall and record group by compound
  dplyr::group_by(across(all_of(group_vars))) %>%
  
  #gets the sum of each compound for each participant's recall
  dplyr::mutate(compound_intake_mg = sum(pp_consumed, na.rm = TRUE)) %>% 
  dplyr::select(c(subject, 
                  any_of(c("RecallNo", "RecordNo", "RecordDayNo")),
                  class, compound_public_id, compound_name, compound_intake_mg, Total_KCAL)) %>%
  dplyr::ungroup()%>%
  
  #Remove duplicates since we've summed each polyphenol per recall
  dplyr::distinct(across(all_of(group_vars)), .keep_all = TRUE) %>%
  
  #Filter out missing compounds, this is for foods that did not map
  dplyr::filter(!is.na(compound_public_id)) %>%
  
  #Standardize Intakes to caloric intake
  dplyr::mutate(compound_intake_mg1000kcal = compound_intake_mg/(Total_KCAL/1000))

# Write output
vroom::vroom_write(compound_intakes_entry,
                   "outputs/summary_compound_intake_by_entry.csv", delim = ",")

Daily Class Intakes by Subject

# First average caloric intakes
kcal_subject = input_kcal %>%
  dplyr::group_by(subject) %>%
  dplyr::summarise(avg_Total_KCAL = mean(Total_KCAL, na.rm = TRUE))

# Then let's average the class intakes
compound_intakes_subject = compound_intakes_entry %>%
  # We will replace these with the subject average
  dplyr::select(-c(Total_KCAL, compound_intake_mg1000kcal)) %>%
  
  #Average polyphenol intake across recalls for each compound
  dplyr::group_by(subject, compound_public_id) %>%
  dplyr::mutate(Avg_compound_intake_mg = mean(compound_intake_mg)) %>%
  dplyr::ungroup() %>%
  
  #Remove duplicates
  dplyr::distinct(subject, compound_public_id, .keep_all = TRUE) %>%
  dplyr::select(-compound_intake_mg) %>%
  
  # Add kcal data
  dplyr::left_join(kcal_subject, by = 'subject') %>%
  
  # Standardize to caloric intake
  dplyr::mutate(compound_intake_mg1000kcal = Avg_compound_intake_mg/(avg_Total_KCAL/1000)) 

# Write Output
vroom::vroom_write(compound_intakes_subject,
                   "outputs/summary_compound_intake_by_subject.csv", delim = ",")

Available for users who prefer a wide format

compound_intakes_subject_wide = compound_intakes_subject %>%
  
  #Transpose dataframe where each column is a participant
  # columns are the compound_public_id for simplicity
  tidyr::pivot_wider(id_cols = subject, names_from = compound_public_id, 
              values_from = compound_intake_mg1000kcal, values_fill = 0)

# Write Output
vroom::vroom_write(compound_intakes_subject_wide,
                   "outputs/summary_compound_intake_by_subject_wide.csv", delim = ",")