R Graphical Manual

Browse All

Last data update: 2014.03.03

R: Normalisation

Normalise

R Documentation

Normalisation

Description

Normalise a metabolomic data matrix according to a specified method.

Usage

Normalise(inputdata, 
    method = c("median","mean","sum","ref", "is", "nomis","ccmn", "ruv2"),
    refvec = NULL, ncomp = NULL, k = NULL, nc = NULL, 
    saveoutput = FALSE, outputname = NULL)

Arguments

`inputdata`	A log transformed data frame in the input data format. See `metabolomics` for details.
`method`	A character string indicating the required normalization method. Must be one of "`median`", "`mean`", "`sum`", "`ref`", "`is`", "`nomis`", "`ccmn`" or "`ruv2`". See Details for information.
`refvec`	A vector of internal standards to be used with the method "`is`", or a reference vector to be used with the method "`ref`".
`ncomp`	Number of PCA components to be used for the "`ccmn`" method. If `NULL`, this will be determined by cross validation as described by Redestig (2012).
`k`	Number of factors of unwanted variation to be included in the "`ruv2`" model.
`nc`	A vector indicating which metabolites should be used as the non-changing metabolites in the "`ruv2`" model, or as multiple internal standards in the "`ccmn`", "`nomis`" and "`ruv2`" methods.
`saveoutput`	A logical indicating whether the normalised data matrix should be saved as a .csv file.
`outputname`	The name of the output file if the output has to be saved.

Details

The normalisation methods based on scaling include normalisation to a total sum, or by the median or mean of each sample, and are denoted by "sum", "median", and "mean" respectively. The method "ref" normalises the metabolite abundances to a specified reference vector.

The normalisation methods based on internal or external standards include "is" which uses a single standard, Cross-contribution Compensating Multiple internal standard Normalisation, "ccmn" (Redestig et al., 2009); normalization using optimal selection of multiple internal standards, "nomis" (Sysi-aho et al. 2007); and "ruv2" (De Livera et al. 2012a).

The Remove Unwanted Varitation "ruv2" method generates a matrix of unwanted variation using non-changing metabolites including any internal or external standards. This matrix of unwanted variation can then be used for identifying differentiallly abundant metabolites in the LinearModelFit function. The RUV2 method attempts to capture both observed and unobserved technical and biological variation (De Livera et al. 2012a, Gagnon-Bartsch et al. 2012).

An overview of these normalisation methods are given by De Livera et al. (2012a, 2012b). Both the "ruv2" and "ccmn" methods use the factors of interest (groups), and therefore should not be used for those unsupervised methods where the groups must be treated as unknown.

Value

The result is an object of class metabdata.

Author(s)

Alysha M De Livera, Jairus B Bowne

References

De Livera, A. M., Dias, D. A., De Souza, D., Rupasinghe, T., Pyke, J., Tull, D., Roessner, U., McConville, M., Speed, T. P. (2012a) Normalising and integrating metabolomics data. Analytical Chemistry 84(24): 1076-10776.

De Livera, A.M., Olshansky, M., Speed, T. P. (2013) Statistical analysis of metabolomics data. Methods in Molecular Biology In press.

Gagnon-Bartsch, Johann A., Speed, T. P. (2012) Using control genes to correct for unwanted variation in microarray data. Biostatistics 13(3): 539-552.

Redestig, H., Fukushima, A., Stenlund, H., Moritz, T., Arita, M., Saito, K., Kusano, M. (2009) Compensation for systematic cross-contribution improves normalization of mass spectrometry based metabolomics data. Analytical Chemistry 81(19): 7974-7980.

Sysi-Aho, M., Katajamaa, M., Yetukuri, L., Oresic, M. (2007) Normalization method for metabolomics data using optimal selection of multiple internal standards. BMC Bioinformatics 8(1): 93.

Examples

    ## Reading the data
    data(mix)
    Y <- log(exprs(mix))
    inputdata <- data.frame(pData(mix)$type, t(Y))
    batch <- pData(mix)$runorder
    nc <- which(with(fData(mix), tag == "IS")==TRUE)
    
    ## Normalise by the median
    norm_med <- Normalise(inputdata, method = "median")
    
    ## Normalise by an internal standard
    norm_is <- Normalise(inputdata, method = "is", 
        refvec=inputdata[, nc[1]])
    
    ## Normalise by a reference vector, in this case an internal standard
    norm_ref <- Normalise(inputdata, method = "ref", 
        refvec = inputdata[, nc[1]])
    
    ## Normalise by the sum
    norm_sum <- Normalise(inputdata, method = "sum")
    
    ## Normalise by the NOMIS method
    norm_nomis <- Normalise(inputdata, method = "nomis", nc = nc)
    
    ## Normalise by the CCMN method
    norm_ccmn <- Normalise(inputdata, method = "ccmn", nc = nc, ncomp = 2)

    ## Normalise using RUV2 method
    norm_ruv2 <- Normalise(inputdata, method = "ruv2", nc = nc, k = 9)

    ## Pca Plots of unwanted variation
    PcaPlots(data.frame(batch, norm_ruv2$output[, -1]), 
        main = "Unwanted batch variation")