sam {siggenes}R Documentation

Significance Analysis of Microarrays

Description

Performs a Significance Analysis of Microarrays (SAM) for a set of positive thresholds. It is possible to do either an one class or a two class SAM analysis.

Usage

sam(data,cl,B=100,balanced=FALSE,mat.samp=NULL,delta=(1:10)/5,med.fdr=TRUE,
        s0=NA,alpha.s0=seq(0,1,.05),include.s0=TRUE,p0=NA,lambda.p0=1,vec.lambda.p0=(0:95)/100,
        na.rm=FALSE,graphic.fdr=TRUE,thres.fdr=seq(0.5,2,0.5),ngenes=NA,iteration=3,
        initial.delta=c(.1,seq(.2,2,.2),4),rand=NA)

Arguments

data the data set that should be analyzed. Every row of this data set must correspond to a gene.
cl a vector containing the class labels of the samples. In the two class unpaired case, the label of a sample is either 0 (e.g., control group) or 1 (e.g., case group). In the two class paired case, the labels are the integers between 1 and n/2 (e.g., before treatment group) and between -1 and -n/2 (e.g., after treatment group), where n is the length of cl and k is paired with -k. For one group data, the label for each sample should be 1.
B number of permutations used in the calculation of the null density. Default is B=100.
balanced if TRUE, balanced permutations will be used. Default is FALSE.
mat.samp a permutation matrix. If specified, this matrix will be used, even if rand and B are specified.
delta a vector of values for the threshold Delta for which the SAM analysis is performed.
med.fdr if TRUE (default), the median number, otherwise the expected number, of falsely called genes will be computed.
s0 the fudge factor. If NA (default), the fudge factor s0 will be computed automatically.
alpha.s0 the possible values of the fudge factor s0 in terms of quantiles of the standard deviations of the genes.
include.s0 if TRUE (default), s0=0 is a possible choice for the fudge factor.
p0 the probability that a gene is not differentially expressed. If not specified (default), it will be computed.
lambda.p0 number between 0 and 1 that is used to estimate p0. If set to 1 (default), the automatic p0 selection using the natural cubic spline fit is used.
vec.lambda.p0 vector of values for λ used in the automatical computation of p0.
na.rm if FALSE (default), the expression scores d of genes with one or more missing values will be set to NA. If TRUE, the missing values will be replaced by the genewise mean of the non-missing values.
graphic.fdr if TRUE (default), both the SAM plot and the plots of Delta vs. FDR and Delta vs. number of significant genes will be generated.
thres.fdr for each value contained in thres.fdr, two lines parallel to the 45-degree line are generated in the SAM plot.
ngenes a number or proportion of genes for which the FDR is estmated.
iteration the number of iterations used in the estimation of the FDR for a given number or proportion of genes.
initial.delta a set of initial guesses for Delta in the computation of the FDR for a given number or proportion of genes.
rand if specified, the random number generator will be put in a reproducible state.

Value

a table of statistics (estimate of p0, number of significant genes, number of falsely called genes and FDR) for the specified set of Deltas, a SAM Plot, a Delta vs. FDR plot, and a plot of Delta vs. the number of significant genes.

Warning

In the one class case, the null distribution will only be computed correctly, if the expression values are log ratios. So in the one class case only log ratios should be used. (There will be no checking, if the expression values are really log ratios.)

Note

For further analyses with sam.plot, the results of sam must be assigned to an object.

SAM was deveoped by Tusher et al. (2001).

!!! There is a patent pending for the SAM technology at Stanford University. !!!

Author(s)

Holger Schwender holger.schw@gmx.de

References

Tusher, V.G., Tibshirani, R., and Chu, G. (2001). Significance analysis of microarrays applied to the ionizing radiation response, PNAS, 98, 5116-5121.

Storey, J.D. (2002). A direct approach to the false discovery rate, Journal of the Royal Statistical Society, Series B, 64, 479-498.

Storey, J.D., and Tibshirani, R. (2003). Statistical significance for genome-wide experiments, Technical Report, Department of Statistics, Stanford University.

Schwender, H. (2003). Assessing the false discovery rate in a statistical analysis of gene expression data, Chapter 5, Diploma thesis, Department of Statistics, University of Dortmund, http://de.geocities.com/holgerschw/thesis.pdf.

See Also

sam.plot sam.wilc sam.delta

Examples

    library(multtest)
    # Load the data of Golub et al. (1999). data(golub) contains a 3051x38 gene expression
    # matrix called golub, a vector of length called golub.cl that consists of the 38 class labels,
    # and a matrix called golub.gnames whose third column contains the gene names.
    data(golub)
    
    # Performing a SAM Analysis of the Golub data. Setting rand=123, to make the results reproducible,
    # and setting med.fdr=FALSE, such that the mean number instead of the median number of falsely called
    # genes is computed. The output is assigned to an object for further analyses.
   if (interactive())  
    sam.output<-sam(golub,golub.cl,med.fdr=FALSE,rand=123)

[Package siggenes version 1.0.6 Index]