twilight.pval {twilight}R Documentation

A function to compute p-values from expression sets

Description

The functions implements twosample t, Z and fold change equivalent tests, paired or unpaired, and correlation coefficients. Based on permutations, expected test statistics as given in Tusher et al. (2001) and empirical p-values are computed. Additional output are q-values computed as given in Storey and Tibshirani (2003). The resulting object is of class twilight and can be passed to functions twilight or plot.twilight.

Usage

twilight.pval(xin, yin, method = "fc", paired = FALSE, B = 1000, yperm = NULL, balance = FALSE, quant.ci = 0.95, s0=NULL, verbose = TRUE)

Arguments

xin Either an expression set (exprSet) or a data matrix with rows corresponding to features and columns corresponding to samples.
yin A numerical vector containing class labels. The higher label denotes the case, the lower label the control samples to test case vs. control. For correlation scores, yin can be any numerical vector of length equal to the number of samples.
method Character string: "fc" for fold change equivalent test (that is log ratio test), "t" for t test, and "z" for Z test. With "pearson" or "spearman", the test statistic is either Pearson's correlation coefficient or Spearman's rank correlation coefficient.
paired Logical value. Depends on whether the samples are paired. Ignored if method="pearson" or method="spearman".
B Numerical value specifying the number of permutations.
yperm Optional matrix containing in each row a permutation of the class labels in binary format. If yperm is specified, no other permutation will be done. Ignored if method="pearson" or method="spearman".
balance Logical value. Depends on whether balanced or unbalanced permutations should be done. Ignored if method="pearson" or method="spearman".
quant.ci Probability value for confidence lines. Lines are symmetric and denote the quant.ci-quantile of maximal absolute differences between each permutatin and the expected scores.
s0 Fudge factor for variance correction in Z test. Takes effect only if method="z". If s0=NULL: Fudge factor is set to median of root pooled variances.
verbose Logical value for message printing.

Details

See vignette.

Value

Returns a twilight object consisting of a data.frame named result with variables

observed Observed test statistics.
expected Mean of order statistics of the permutation statistics.
candidate Binary vector. "1" for observations exceeding the confidence lines.
pvalue Twosided test permutation p-values.
qvalue q-values computed as described in Storey and Tibshirani (2003).
index Index of the original ordering.

Values are sorted by absolute observed scores.

Note

Additional output consists of

ci.line Quantile corresponding to quant.ci, passed for plotting.
pi0 Estimated prior probability.
call Character string of function arguments.
quant.ci Passes quant.ci for plotting.

The remaining slots are left free for function twilight.

Author(s)

Stefanie Scheid http://www.molgen.mpg.de/~scheid

References

Golub TR, Slonim DK, Tamayo P, Huard C, Gaasenbeek M, Mesirov JP, Coller H, Loh ML, Downing JR, Caligiuri MA, Bloomfield CD and Lander ES (1999): Molecular Classification of Cancer: Class Discovery and Class Prediction by Gene Expression Monitoring, Science 286, 531–537.

Huber W, von Heydebreck A, Sültmann H, Poustka A and Vingron M (2002): Variance stabilization applied to microarray data calibration and to the quantification of differential expression, Bioinformatics 18, suppl. 1, S96–S104.

Scheid S and Spang R (2004): A stochastic downhill search algorithm for estimating the local false discovery rate, IEEE TCBB 1(3), 98–108.

Storey JD and Tibshirani R (2003): Statistical significance for genomewide studies, PNAS 100(16), 9440–9445.

Tusher VG, Tibshirani R and Chu G (2001): Significance analysis of mircroarrays applied to the ionizing response, PNAS 98(9), 5116–5121.

See Also

twilight, plot.twilight, twilight.combi

Examples

### Leukemia data set of Golub et al. (1999)
library(golubEsets)
data(golubMerge)

### Variance-stabilizing normalization of Huber et al. (2002)
library(vsn)
golubNorm <- vsn(exprs(golubMerge))

### A vector of class labels.
id <- as.numeric(golubMerge$ALL.AML)

a <- twilight.pval(golubNorm,id)
plot(a,which="scores")
plot(a,which="qvalues")

[Package twilight version 1.1.0 Index]