reducing: Grouping chemically redundant MS1 features

In SciDoPhenIA/phenomis: Analysis of single and multiple phenomic data sets

Grouping chemically redundant MS1 features

Description

This method groups chemically redundant features from a peak table, based on 1) correlation of sample profiles, 2) retention time window, 3) referenced m/z differences. The initial algorithm is named 'Analytic Correlation Filtration' (Monnerie et al., 2019; DOI:10.3390/metabo9110250) and is available in Perl and on the Workflow4Metabolomics platform. Here, the algorithm described in the paper was implemented in R as follows: An adjacency matrix of all pairs of features is built, containing a 1 when the features have a (Pearson) correlation above the (0.9) threshold, a retention time difference between the (6) seconds threshold, and an m/z difference belonging to referenced adducts, isotopes and fragments m/z difference, and containing a 0 otherwise. The connex components of this adjacency matrix are extracted ('igraph' package). Within each component, the features are ranked by decreasing average intensity in samples; all features except the first one are flagged as 'redundant'. Note: the algorithm relies on the 'mzdiff_db.tsv' file referencing the known adducts, isotopes, and fragments.

Usage

reducing(
  x,
  cor_method.c = "pearson",
  cor_threshold.n = 0.9,
  rt_tol.n = 6,
  rt_colname.c = "rt",
  mzdiff_tol.n = 0.005,
  mz_colname.c = "mz",
  return_adjacency.l = FALSE,
  report.c = c("none", "interactive", "myfile.txt")[2]
)

## S4 method for signature 'MultiAssayExperiment'
reducing(
  x,
  cor_method.c = "pearson",
  cor_threshold.n = 0.9,
  rt_tol.n = 6,
  rt_colname.c = "rt",
  mzdiff_tol.n = 0.005,
  mz_colname.c = "mz",
  return_adjacency.l = FALSE,
  report.c = c("none", "interactive", "myfile.txt")[2]
)

## S4 method for signature 'SummarizedExperiment'
reducing(
  x,
  cor_method.c = "pearson",
  cor_threshold.n = 0.9,
  rt_tol.n = 6,
  rt_colname.c = "rt",
  mzdiff_tol.n = 0.005,
  mz_colname.c = "mz",
  return_adjacency.l = FALSE,
  report.c = c("none", "interactive", "myfile.txt")[2]
)

## S4 method for signature 'MultiDataSet'
reducing(
  x,
  cor_method.c = "pearson",
  cor_threshold.n = 0.9,
  rt_tol.n = 6,
  rt_colname.c = "rt",
  mzdiff_tol.n = 0.005,
  mz_colname.c = "mz",
  return_adjacency.l = FALSE,
  report.c = c("none", "interactive", "myfile.txt")[2]
)

## S4 method for signature 'ExpressionSet'
reducing(
  x,
  cor_method.c = "pearson",
  cor_threshold.n = 0.9,
  rt_tol.n = 6,
  rt_colname.c = "rt",
  mzdiff_tol.n = 0.005,
  mz_colname.c = "mz",
  return_adjacency.l = FALSE,
  report.c = c("none", "interactive", "myfile.txt")[2]
)

Arguments

x	An S4 object of class SummarizedExperiment or MultiAssayExperiment: the dataset(s) must contain the dataMatrix and the variableMetadata (with the 'mz' and 'rt' columns)
cor_method.c	character(1): correlation method (default: 'pearson')
cor_threshold.n	numeric(1): correlation threshold (default: 0.9)
rt_tol.n	numeric(1): retention time width in seconds (default: 6 s)
rt_colname.c	character(1): column name for the retention time in the fData (default: 'rt')
mzdiff_tol.n	numeric(1): tolerance in Da for the matching of m/z differences and referenced adducts, isotopes, and fragments (default: 0.005 Da)
mz_colname.c	character(1): column name for the m/z in the fData (default: 'mz')
return_adjacency.l	logical(1): should the adjacency matrix be returned (in addition to the updated ExpressionSet)?
report.c	character(1): File name with '.txt' extension for the printed results (call to sink()'); if 'interactive' (default), messages will be printed on the screen; if 'none', no verbose will be generated

Value

updated SummarizedExperiment or MultiAssayExperiment: the summarized experiment(s) now include(s) 5 new columns in the rowData: 'redund_samp_mean', 'redund_is', 'redund_group', redund_iso_add_frag', 'redund_repres' and 'redund_relative' containing, respectively, the redundant features (coded by 1; i.e. features with a relative annotation distinct from ” and 'M'), the connected components, the m/z diff. chemical annotations, the representative ion of each group, and the annotations relative to this representative ion within each group

Examples

sacurine.se <- reading(system.file("extdata/W4M00002_Sacurine-comprehensive",
                                          package = "phenomis"),
                      report.c = "none")
sacurine.se <- reducing(sacurine.se,
                        rt_tol.n = 6)
table(rowData(sacurine.se)[, "redund_group"])

SciDoPhenIA/phenomis documentation built on June 9, 2022, 11:54 p.m.