loessSmooth: pooled quality control based signal drift/ attenuation...

In WMBEdmands/MetMSLine: an automated and fully integrated pipeline for rapid processing of high-resolution LC-MS metabolomic datasets.

Description

attempts to correct for systematic drifts in LC-MS datasets using intervally injected pooled quality control samples representative of all samples of a experiment.

Usage

loessSmooth(peakTable = NULL, sampNames = NULL, qcNames = NULL,
  nCores = NULL, outputDir = NULL, smoothSpan = NULL, folds = 7)

Arguments

peakTable	either a data.frame, full file path as a character string to a .csv file of a peak table in the form observation (samples) in columns and variables (Mass spectral signals) in rows. If argument is not supplied a GUI file selection window will open and a .csv file can be selected.
sampNames	character vector of sample names to identify appropriate observation (sample) columns.
qcNames	character vector of quality control (QC) names to identify appropriate observation (sample) columns.
nCores	number of cores to use for parallel processing with the foreach and snow packages.
outputDir	optional directory path to save output images before and after QC smoothing. A subdirectory will be created in which to save the png images.
smoothSpan	fixed smoothing span. If supplied a this fixed smoothing span parameter will override the cross validated feature-by-feature smoothing span optimization.
folds	numeric (default=7, i.e. 7-fold cross validation).

Details

systematic drifts in mass spectral signal can be corrected using intervally injected pooled quality control samples for each mass spectral signal variable in the peak table. The optimum span parameter for each LC-MS feature is identified using 7-fold cross-validation (see: crossval from the bootstrap package). A new column is added to the peakTable "smoothSpanLoessFit" containing the result of the optimal value of span for the loess function. For each LC-MS variable loess models of the quality control data are calculated using direct surface fitting (see: loess.control) and the loess fit using 7-fold cross validation is carried out by a slight modification of the loess.wrapper function from the bisoreg package. The optimum loess model is then used to predict (see: predict.loess) intensity values for both the QC and sample injections according to the degree of smoothing (span) in the final optimum loess model. The product of the original LC-MS variable intensity and the deviation of the loess predicted values from their median is then calculated to adjust the signal drift. This function attempts to adjust systematic drift/ attenuation within each LC-MS feature. The degree of smoothing determined by the smoothSpan argument however results may vary slightly using the K-fold cross validation method. In order to maintain absolute reproducibility it is recommend to set the seed prior to using this function (see: set.seed). It is recommended that parallel processing is considered to reduce the computational time cost using the argument nCores.

Value

a data frame identical to peakTable with signal drift/ attenuation adjusted.

Source

http://www.nature.com/nprot/journal/v6/n7/abs/nprot.2011.335.html Procedures for large-scale metabolic profiling of serum and plasma using gas chromatography and liquid chromatography coupled to mass spectrometry

See Also

loess, loess.wrapper, bisoreg, crossval, predict.loess, set.seed.

WMBEdmands/MetMSLine documentation built on May 9, 2019, 10:03 p.m.