Discovering regulated Metabolite Families

Abstract

Description of the MetFamily R package

MetFamily

MetFamily is an R package that provides a novel approach for the untargeted discovery of metabolite families.

Read more on the GitHub repository: https://github.com/ipb-halle/MetFamily

We first need to install the development version of the package.

remotes::install_github("ipb-halle/MetFamily@devel")

# And load it
library(MetFamily)

Dataset

The following analysis looks into metabolites found in glandular trichomes of tomato plants. See the referenced paper for additional information [1].

The dataset is provided in the MetFamily package.

# MS1 intensities
filePeakMatrixPath <- system.file("extdata/showcase/Metabolite_profile_showcase.txt", package = "MetFamily")

# MS2 spectra
fileSpectra <- system.file("extdata/showcase/MSMS_library_showcase.msp", package = "MetFamily")

# Sirius annotations
fileAnnotation <- system.file("extdata/testdata/canopus/canopus1680.txt", package = "MetFamily")

# We use default import parameters
parameterSet <- parameterSetDefault()
parameterSet$minimumIntensityOfMaximalMS2peak <- 2000
parameterSet$minimumProportionOfMS2peaks <- 0.05


dataList <- projectFromFiles(
  ms1_path = filePeakMatrixPath, 
  ms2_path = fileSpectra, 
  siriusFileColumnName = "ClassyFire class",
  parameterSet = parameterSet,
  annot_path = fileAnnotation
)

## [1] "Parsing MS/MS file..."
## [1] "Parsing MS/MS file..."

## MS/MS file: Read file

## MS/MS file: Parse

## MS/MS file: Filter

## MS/MS file: Assemble spectra

## [1] "Parsing MS/MS file ready"
## [1] "Parsing MS1 file..."
## [1] "Parsing MS1 file content..."
## [1] "Precursor deisotoping..."
## [1] "Precursor deisotoping 582 / 5823"
## [1] "Precursor deisotoping 1164 / 5823"
## [1] "Precursor deisotoping 1746 / 5823"
## [1] "Precursor deisotoping 2328 / 5823"
## [1] "Precursor deisotoping 2910 / 5823"
## [1] "Precursor deisotoping 3492 / 5823"
## [1] "Precursor deisotoping 4074 / 5823"
## [1] "Precursor deisotoping 4656 / 5823"
## [1] "Precursor deisotoping 5238 / 5823"
## [1] "Precursor deisotoping 5820 / 5823"
## [1] "Boxing..."
## [1] "Postprocessing matrix..."
## [1] "Building fragment mzFragmentGroups..."
## [1] "Fragment grouping preprocessing..."
## [1] "Fragment grouping preprocessing ready"
## [1] "Fragment grouping"
## [1] "Fragment grouping 34 / 86970"
## [1] "Fragment grouping 27083 / 86970"
## 
## [1] "Fragment grouping ready (4.00752353668213s)"
## [1] "Fragment group postprocessing"
## [1] "Fragment group postprocessing: 1858 / 18589"
## [1] "Fragment group postprocessing: 3716 / 18589"
## [1] "Fragment group postprocessing: 5574 / 18589"
## [1] "Fragment group postprocessing: 7432 / 18589"
## [1] "Fragment group postprocessing: 9290 / 18589"
## [1] "Fragment group postprocessing: 11148 / 18589"
## [1] "Fragment group postprocessing: 13006 / 18589"
## [1] "Fragment group postprocessing: 14864 / 18589"
## [1] "Fragment group postprocessing: 16722 / 18589"
## [1] "Fragment group postprocessing: 18580 / 18589"
## [1] "Fragment group postprocessing ready (2.77388501167297s)"
## [1] "Boxing to matrix"
## [1] "Fragment group deisotoping"
## [1] "Fragment group deisotoping 1858 / 18589"
## [1] "Fragment group deisotoping 3716 / 18589"
## [1] "Fragment group deisotoping 5574 / 18589"
## [1] "Fragment group deisotoping 7432 / 18589"
## [1] "Fragment group deisotoping 9290 / 18589"
## [1] "Fragment group deisotoping 11148 / 18589"
## [1] "Fragment group deisotoping 13006 / 18589"
## [1] "Fragment group deisotoping 14864 / 18589"
## [1] "Fragment group deisotoping 16722 / 18589"
## [1] "Fragment group deisotoping 18580 / 18589"
## [1] "Fragment group deisotoping ready (13.2282903194427s)"
## [1] "Fragment group boxing"
## [1] "Building fragment mzFragmentGroups ready"

## 'as(<dgCMatrix>, "dgTMatrix")' is deprecated.
## Use 'as(., "TsparseMatrix")' instead.
## See help("Deprecated") and help("Matrix-deprecated").

## [1] "Boxing..."
## [1] "Ready"
## [1] "Preprocessing 473 / 2463"
## [1] "Preprocessing 1222 / 2463"
## [1] "Preprocessing 1965 / 2463"
## [1] "Features"
## [1] "Feature postprocessing"
## [1] "Coloring"
## [1] "Coloring init"
## [1] "Coloring naming functions"
## [1] "Coloring gather data"
## [1] "Coloring matrix"
## [1] "Feature annotations"
## [1] "Boxing"
## [1] "Ready"
## [1] "Merging by: featureId and Alignment ID"

Filtering data

We use a filter for the intensity threshold and log2-fold-change to remove noise and focus on features of interest.

filterObj <- makeFilterObj(dataList, filter_average = 10000, filter_lfc = 2)

PCA figures

pca <- calculatePCA(dataList=dataList,
                    filterObj=filterObj, 
                    ms1AnalysisMethod="PCA (Principal Component Analysis)", 
                    scaling="None", 
                    logTransform=FALSE)

## [1] "######################################################################################"
## [1] "PCA (Principal Component Analysis)"
## [1] "Analysis: pcaMethods"

resultObj <- calcPlotPCAscores(
  pcaObj = pca, 
  dataList = dataList,
  filterObj = filterObj,
  showScoresLabels = FALSE, 
  xInterval = NULL, 
  yInterval = NULL
)

PCA scores of MS1

resultObj <- calcPlotPCAloadings(
  pcaObj = pca,
  dataList = dataList,
  # filter = filterObj$filter, # deprecated
  selectionFragmentPcaLoadingSet = NULL,
  selectionAnalysisPcaLoadingSet = NULL,
  selectionSearchPcaLoadingSet   = NULL,
  xInterval = NULL,
  yInterval = NULL,
  loadingsLabels = "None",
  showLoadingsAbundance = FALSE,
  showLoadingsFeaturesAnnotated   = TRUE,
  showLoadingsFeaturesUnannotated = TRUE,
  showLoadingsFeaturesSelected    = TRUE,
  showLoadingsFeaturesUnselected  = TRUE
)

PCA loadings of MS1

HCA figure on MS2

In construction…

fileName <- system.file("extdata/testdata/clusterDataList.Rdata", package = "MetFamily")
load(fileName) 
fileName <- system.file("extdata/testdata/hcaFilter.Rdata", package = "MetFamily")
load(fileName) 

tryCatch(
  returnObj <- calcPlotDendrogram(
    dataList=dataList, #project, 
    filter=filter, 
    clusterDataList=clusterDataList, 
    annoPresentAnnotationsList = annoPresentAnnotationsList ,
    annoPresentColorsList = annoPresentColorsList,
    distanceMeasure="Jaccard (intensity-weighted)", 
    selectionFragmentTreeNodeSet = NULL, 
    selectionAnalysisTreeNodeSet = NULL, 
    selectionSearchTreeNodeSet = NULL, 
    showClusterLabels = TRUE, 
    hcaPrecursorLabels = "m/z / RT", 
    xInterval = c(1,219)),
  error = \(x){invisible(NULL)}
)

HCA on MS2

References

1. Treutler H, Tsugawa H, Porzel A, Gorzolka K, Tissier A, Neumann S, Balcke GU: Discovering Regulated Metabolite Families in Untargeted Metabolomics Studies. Anal Chem 2016.

Steffen Neumann

Guillaume Patoine