RIcorrect: Peak picking from CDF files and RI correction

In acinostroza/TargetSearch: A package for the analysis of GC-MS metabolite profiling data

Peak picking from CDF files and RI correction

Description

This function reads from CDF files, finds the apex intensities, converts the retention time to retention time index (RI), and writes RI corrected text files (a.k.a. RI files). In addition, it can perform baseline correction and also convert files to the new NetCDF-4 TargetSearch format.

Usage

RIcorrect(samples, rimLimits = NULL, massRange=NULL, Window, IntThreshold,
         pp.method="ppc", showProgressBar=FALSE, baseline=FALSE,
         writeCDF4path=TRUE, ...)

Arguments

samples	A tsSample object usually created by ImportSamples function.
rimLimits	A tsRim object. If set to NULL, no retention time will be performed. See ImportFameSettings.
massRange	Deprecated. It is completely ignored but it is kept for compatibility with old scripts.
Window	The window used for smoothing. The number of points actually used is 2*Window + 1. It must be an integer. See details.
IntThreshold	Apex intensities lower than this value will be removed from the RI files.
pp.method	Peak picking method. Options are "smoothing", "gaussian" and "ppc". See details.
showProgressBar	Logical. Should the progress bar be displayed?
baseline	Logical. Should baseline correction be performed?
writeCDF4path	Whether or not convert a CDF-3 into a CDF-4. It can take a logical value or a character vector representing file paths. See details below.
...	A list of options passed to baseline.

Details

There are three pick picking methods available: "ppc", "smoothing", "gaussian".

The "ppc" method (default) implements the peak detection method described in the ppc package. It looks for the local maxima within a 2*Window + 1 scans for every mass trace.

The "smoothing" method calculates a moving average of 2*Window + 1 points for every mass trace. Then it looks for a change of sign (from positive to negative) of the difference between two consecutive points. Those points will be returned as detected peaks.

The "gaussian" method behaves similar to the "smoothing" method, but instead a gaussian smoother is used instead of the moving average.

To work out a suitable Window value, the following might be useful: Window = (SR * PW - 1) / 2, where SR is the scan rate of the MS instrument and PW is the peak width. Because Window is an integer, the resulting value must be rounded. For example, for SR = 20 scans per second, a PW = 1.5 seconds, then Window = 14.5, which can be rounded to 15.

The RI file type is determined by the output of fileFormat method applied to the tsSample input object. To choose between the available formats ("binary" and "text"), select it with fileFormat method before calling RIcorrect.

The parameter writeCDF4path is used to convert CDF-3 files into a custom CDF-4 format. It can be logical or a character vector. If TRUE, the default, the CDF-3 files will be converted automatically to CDF-4 format. The files will be saved in the same directory as the original CDF-3. If it is character vector representing file paths, then the CDF-4 will be saved in those paths instead (re-cycled to the length of samples). Finally, if FALSE, then no CDF conversion will be performed. This is not recommended, but if possible if you want to match the old TargetSearch behaviour. Note that also the baseline correction or the retention time indices will not be updated as well.

If baseline is TRUE, the CDF files will be baseline corrected by passing the extra parameters to baseline. See there for details.

Value

A retention time matrix of the found retention time markers. Every column represents a sample and rows RT markers.

Author(s)

Alvaro Cuadros-Inostroza, Matthew Hannah, Henning Redestig

See Also

ImportSamples, ImportFameSettings, NetCDFPeakFinding, FAMEoutliers, tsSample, tsRim.

Examples

require(TargetSearchData)
# import refLibrary, rimLimits and sampleDescription.
data(TSExample)
# get the CDF files
cdfpath <- tsd_data_path()
cdfpath
list.files(cdfpath)
# update the CDF path
CDFpath(sampleDescription) <- cdfpath

# change file format of RI files as bin
fileFormat(sampleDescription) <- 'binary'

# Parameters: Intensity Threshold = 50 peak detection method = "ppc", window = 15
# To match the old behavior, the do not create CDF-4 Files (not recommended)
RImatrix <- RIcorrect(sampleDescription, rimLimits, writeCDF4path=FALSE,
            Window = 15, pp.method = "ppc", IntThreshold = 50)

# Convert to CDF-4 (recommended) with same parameters
# Note: save files in same directory (as
RImatrix <- RIcorrect(sampleDescription, rimLimits, writeCDF4path=".",
            Window = 15, pp.method = "ppc", IntThreshold = 50)

# we need to update the sampleDescription to use the new files
# this is not done automatically
sampleDescription <- ncdf4Convert(sampleDescription, ".")

# you can try other parameters and other peak picking algorithm
RImatrix <- RIcorrect(sampleDescription, rimLimits,
            Window = 15, pp.method = "smoothing", IntThreshold = 10)

RImatrix <- RIcorrect(sampleDescription, rimLimits,
            Window = 15, pp.method = "ppc", IntThreshold = 100)

acinostroza/TargetSearch documentation built on Nov. 22, 2024, 3:31 p.m.