data_transformation.md

recetox-aplcms to xMSannotator

The recetox-aplcms tool outputs the peak table in a format which is not directly compatible with recetox-xMSannotator or xMSannotator. A conversion is required. Example file layouts are given below.

recetox-aplcms Example Output

feature	mz	rt	sample	sample_rt	sample_intensity
1	110.034534364169	246.052915606977	Tribrid_201005_001-MeOH_NEG_MU	NA	0
1	110.034534364169	246.052915606977	Tribrid_201005_002-Norman_NEG_MU	729.439375046056	256906.674516574
2	113.007573530083	370.352087874597	Tribrid_201005_001-MeOH_NEG_MU	317.947801280858	2406770.19370756
...	...	...	...	...	...

recetox-xMSannotator Example Input

peak	mz	rt	Tribrid_201005_001-MeOH_NEG_MU	Tribrid_201005_002-Norman_NEG_MU
1	110.034534364169	246.052915606977	0	256906.674516574
...	...	...	...	...

xMSannotator Example Input

mz	time	intensity_Tribrid_201005_001-MeOH_NEG_MU	intensity_Tribrid_201005_002-Norman_NEG_MU
110.034534364169	246.052915606977	0	256906.674516574
...	...	...	...

R Sample Code

#' @param file Path to parquet file to be loaded into memory
#'
#' @export
load_parquet <- function(file) {
    message <- paste("The file", toString(file), "seams not to be a valid Parquet file.")
    rlang::with_handlers(
        arrow::read_parquet(file),
        error = ~ rlang::abort(message, parent = .)
    )
}


#' @param peak_table Peak table from recetox.aplcms package.
#'
#' @export
#' @import dplyr
rcx_aplcms_to_xmsannotator <- function(peak_table) {
    col_base <- c("feature", "mz", "rt")
    output_table <- peak_table %>% distinct(across(any_of(col_base)))

    for (level in levels(peak_table$sample)) {
        subdata <- peak_table %>%
            filter(sample == level) %>%
            select(any_of(c(col_base, "sample_intensity"))) %>%
            rename(!!paste0("intensity_", level) := "sample_intensity")
        output_table <- inner_join(output_table, subdata, by = col_base)
    }

    output_table <- output_table  %>% select(-feature) %>% rename(time = rt)
    return(output_table)
}

recetox-aplcms to recetox-xMSannotator

This is the same as to the recetox-xMSannotator, but the feature column has to be renamed to peak, and rt can stay the same.

R Sample Code

#' @param peak_table Peak table from recetox.aplcms package.
#'
#' @export
#' @import dplyr
rcx_aplcms_to_rcx_xmsannotator <- function(peak_table) {
    col_base <- c("feature", "mz", "rt")
    output_table <- peak_table %>% distinct(across(any_of(col_base)))

    for (level in levels(peak_table$sample)) {
        subdata <- peak_table %>%
            filter(sample == level) %>%
            select(any_of(c(col_base, "sample_intensity"))) %>%
            rename(!!level := "sample_intensity")
        output_table <- inner_join(output_table, subdata, by = col_base)
    }

    output_table <- output_table %>% rename(peak = feature)
    return(output_table)
}

HMDB: Rda to Parquet

These are simple instructions how to convert RDA file containing the HMDB annotation database (used in xMSannotator) to parquet format.

The important steps are defining filter_keys and target_keys, which are vectors containing keywords to be selected and subsequently renamed, in the respective order.

R Sample Code

save_parquet <- function(data, filename) {
  invisible(arrow::write_parquet(data, filename))
}

load_rda <- function(filename) {
  get(load(filename))
}

convert_rda_to_parquet <- function(rda, filter_keys, target_keys) {
  target_data <- rda[filter_keys]
  names(target_data) <- target_keys
  return(target_data)
}

rda <- load_rda("filename.rda")
filter_keys <- c("Name", "InChiKey", "Formula", "MonoisotopicMass")
target_keys <- c("iupac_name", "iupac_inchikey", "molecular_formula", "monoisotopic_mass")
result <- convert_rda_to_parquet(rda, filter_keys, target_keys)
save_parquet(result, "filename.parquet")