preproc: Data Tranformation Wrapper

Description Usage Arguments Details Value Author(s) References Examples

View source: R/preproc.R

Description

Wrapper for several techniques to perform data transformation of the original fingerprint matrix.

Usage

1
  preproc (x, y=NULL,method="log",add=1)

Arguments

x

A numeric data frame or matrix to be pre-processed.

method

A method used to pre-process the data set. The following methods are supported:

  • center: Centering

  • auto: Auto scaling

  • range: Range scaling

  • pareto: Pareto scaling

  • vast: Vast scaling

  • level: Level scaling

  • log: Log transformation (default)

  • log10: Log 10 transformation

  • sqrt: Square root transformation

  • asinh: Inverse hyperbolic sine transformation

  • TICnorm: TIC normalisation

add

A numeric value for addition used in the logarithmic transformations log and log10.

y

A factor specifying the class for each observation. It is only used by the method TICnorm.

Details

Purpose of normalisation is to remove inter-spectrum sources of variability that come mainly from different sample concentration, loss of sensitivity of the detector over time or degradation of certain samples. Global normalisation by rescaling each measurement within a spectrum by a constant factor, such as the sum of all the spectra intensities (TICnorm method). However, the default use of TIC normalisation can lead to the generation of spurious knowledge if the overall sample intensity is class/factor dependent. If this is the case, a straightforward approach is to remove the difference, between the average scale of the corresponding class and the average scale, from the scale factor by providing the class/factor of interest (argument y) while calling preproc.

Value

Transformed sample matrix.

Author(s)

Wanchang Lin wll@aber.ac.uk and David Enot dle@aber.ac.uk.

References

Berg, R., Hoefsloot, H., Westerhuis, J., Smilde, A. and Werf, M. (2006), Centering, scaling, and transformations: improving the biological information content of metabolomics data, BMC Genomics, 7:142

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
data(abr1)
cl   <- factor(abr1$fact$class)
dat  <- abr1$pos[,110:2000]

## normalise data set using class dependent "TICnorm"
z.1  <- preproc(dat, y=cl, method="TICnorm")

## scale data set using "log10"
z.2 <- preproc(dat,method="log10", add=1)

## or scale data set using "log" and "TICnorm" sequentially
z.3 <- preproc(dat,method=c("log","TICnorm"), add=0.1)

aberHRML/FIEmspro documentation built on May 16, 2019, 6:56 p.m.