Conjugate Screening

This vignette explains how to use the metamorphr package to search a featuretable for characteristic MS/MS fragments and neutral losses.

Introduction

Glutathion (GSH) conjugates of xenobiotics usually result in characteristic MS/MS fragments and neutral losses (Brink et al., 2014). This vignette shows you two approaches, one relying on MS/MS fragments and one relying on neutral losses, to screen a metabolomics data set with associated MS/MS spectra for GSH conjugates. If you want to learn more about drug metabolism you can visit the Wikipedia page (https://en.wikipedia.org/wiki/Drug_metabolism). The data set was obtained by treating HepG2 cells with the redox cycler menadione and measuring the samples in ESI positive mode.

Prepare the data set

This vignette uses data from a separate GitHub repository. More information about the data can be found there.

First, the featuretable, the sample metadata, and the MS/MS spectra have to be imported and joined together. The metamorphr package provides several convenience functions to help with these tasks. To read the sample metadata in the form of a .csv file, readr::read_csv can be used.

menadione_ft <- read_featuretable(
  "https://github.com/yasche/metamorphr-data/raw/refs/heads/main/RP18/pos/MetaboScape/menadione/menadione.csv", 
  label_col = 3, 
  metadata_cols = 1:15)

menadione_metadata <- readr::read_csv(
  "https://github.com/yasche/metamorphr-data/raw/refs/heads/main/RP18/pos/MetaboScape/menadione/menadione_metadata.csv", 
  show_col_types = FALSE)

menadione_msn <- read_mgf("https://github.com/yasche/metamorphr-data/raw/refs/heads/main/RP18/pos/MetaboScape/menadione/menadione.gnps.mgf",
                          show_progress = F)

After the files have been imported successfully, they can be joined together. First the sample metadata is joined using the convenience function join_metadata. The MS/MS spectra can simply be added via dplyr::left_join.

menadione_ft <- menadione_ft %>%
  join_metadata(menadione_metadata) %>%
  dplyr::left_join(menadione_msn, by = "FEATURE_ID")

head(menadione_ft)
#> # A tibble: 6 × 30
#>     UID Feature    Sample           Intensity FEATURE_ID    RT   CCS SIGMA_SCORE
#>   <int> <chr>      <chr>                <dbl>      <dbl> <dbl> <dbl>       <dbl>
#> 1     1 186.015    QC_1_07-Aug-24_…      745.          1  13.6  133.        NA  
#> 2     2 837.83232  QC_1_07-Aug-24_…     1256.          2  24.2  152.        NA  
#> 3     3 769.84306  QC_1_07-Aug-24_…     3594.          3  24.6  223.        NA  
#> 4     4 1403.71268 QC_1_07-Aug-24_…       NA           4  24.7  341.        NA  
#> 5     5 1539.68908 QC_1_07-Aug-24_…       NA           5  25    354.        NA  
#> 6     6 244.22612  QC_1_07-Aug-24_…    25572.          6  25.9  157.        12.9
#> # ℹ 22 more variables: NAME_METABOSCAPE <chr>, MOLECULAR_FORMULA <chr>,
#> #   ADDUCT <chr>, KEGG <chr>, CAS <chr>, MaxIntensity <dbl>,
#> #   Condition_QC_MeanIntensity <dbl>, Condition_BLANK_MeanIntensity <dbl>,
#> #   Condition_Control_MeanIntensity <dbl>,
#> #   Condition_Treatment_MeanIntensity <dbl>, Group <chr>, Replicate <dbl>,
#> #   Batch <dbl>, Factor <dbl>, SCANS <dbl>, PEPMASS <dbl>, MSLEVEL <dbl>,
#> #   CHARGE <chr>, POLARITY <chr>, RTINMINUTES <dbl>, ION <chr>, MSn <list>

Find potential glutathion conjugates

To find potential GSH conjugates, two approaches can be used: (i) search for specific MS/MS spectra and (ii) search for specific neutral losses. Functions for both approaches are implemented in the metamorphr package and are shown below. Results can be easily visualized using the ‘ggplot2’ package. Characteristic fragments and neutral losses are described in Brink et al., 2014.

Some initial filtering

To reduce the search space and remove potential false-positives, some initial filtering can be done using one or more of the several filter functions provided by the metamorphr package. First, QC samples are removed using dplyr::filter as they are not needed for further analysis. In this scenario, we are only interested in features that are solely present in the treated samples, so filter_blank can be used prior to searching for specific fragments or neutral losses. Additionally, we are only interested in features that are found reproducibly across samples, so we also use filter_grouped_mv. Finally, blank samples are removed, again using dplyr::filter.

menadione_ft <- menadione_ft %>%
  dplyr::filter(Group != "QC") %>%
  filter_grouped_mv(min_found = 0.75)  %>%
  filter_blank(blank_samples = c("blank", "control"),
               blank_as_group = T,
               group_column = Group,
               min_frac = 3) %>%
  dplyr::filter(Group != "blank")

Characteristic MS/MS fragments

Searching for characteristic MS/MS fragments can be done using the filter_msn function.

fragments <- c(308.0911,
               179.0485,
               177.0328,
               162.0219,
               130.0499)

menadione_ft %>%
  filter_msn(fragments = fragments, 
             min_found = 2, 
             tolerance = 5, 
             tolerance_type = "ppm", 
             show_progress = F) %>%
  ggplot2::ggplot(ggplot2::aes(Group, Intensity)) +
    ggplot2::geom_point(position = "jitter") +
    ggplot2::facet_wrap(~ Feature) +
    ggplot2::ylim(c(0, NA))
#> Warning: Removed 12 rows containing missing values or values outside the scale range
#> (`geom_point()`).

Searching for characteristic MS/MS fragments yields only one potential GSH conjugate: A feature with an exact mass of 511.12575 Da.

Characteristic neutral losses

Prior the searching for specific neutral losses via the filter_neutral_loss function, neutral loss spectra must be calculated using calc_neutral_loss. The usage of both functions is shown below.

losses <- c(75.0320,
            129.0426,
            232.0696,
            146.0692,
            249.0961,
            273.0961,
            275.1117,
            305.0682,
            307.0838)

menadione_ft %>%
  calc_neutral_loss(m_z_col = PEPMASS) %>%
  filter_neutral_loss(losses = losses, 
                      min_found = 4, 
                      tolerance = 10, 
                      tolerance_type = "ppm",
                      show_progress = F) %>%
  ggplot2::ggplot(ggplot2::aes(Group, Intensity)) +
    ggplot2::geom_point(position = "jitter") +
    ggplot2::facet_wrap(~ Feature) +
    ggplot2::ylim(c(0, NA))
#> Warning: Removed 24 rows containing missing values or values outside the scale range
#> (`geom_point()`).

This approach yields two potential GSH conjugates: One with an exact mass of 477.12018 and the one with an exact mass of 511.12575 which was also found using the MS/MS fragments-based approach above.

Interestingly, menadione has been reported to form a GSH conjugate, thiodione, with a calculated exact mass of 477.12058 Da (Mauzeroll et al., 2004). The difference between the measured exact mass of one potential conjugate and the calculated mass is just 0.0004 Da or 0.86 ppm.