ROTS: Reproducibility-Optimized Test Statistic (ROTS)

Description

Calculates the reproducibility-optimized test statistic (ROTS) for ranking genes in order of evidence for differential expression in two-group comparisons.

Usage

ROTS(data, groups, B, K, paired = FALSE, seed = NULL, a1 = NULL, a2 = NULL, log = TRUE, progress = FALSE)

Arguments

data

a numeric data matrix or an ExpressionSet instance, in which rows correspond to genes and columns correspond to samples.

groups

a vector indicating the sample groups.

an integer specifying the number of bootstrap and permutation resamplings.

an integer indicating the largest top list size considered.

paired

a logical indicating whether a paired test is performed.

seed

an integer seed for the random number generator.

a1, a2

a non-negative parameter. See details section for further information.

log

a logical indicating whether input data is log2 scaled.

progress

a logical indicating if additional progress bars are shown.

Value

data: the expression data matrix.
B: the number of bootstrap and permutation resamplings.
d: the value of the optimized ROTS-statistic for each gene.
pvalue: the corresponding pvalues.
FDR: the corresponding FDR-values.
a1: the optimized parameter a1.
a2: the optimized parameter a2.
k: the optimized top list size.
R: the optimized reproducibility value.
Z: the optimized reproducibility Z-score.

Details

The reproducibility-optimization procedure ROTS enables the selection of a suitable gene ranking statistic directly from the given dataset. The statistic is optimized among a family of t-type statistics d = m/(a1+a2*s), where m is the absolute difference between the group averages, s is the pooled standard error, and a1 and a2 are the non-negative parameters to be optimized. Two special cases of this family are the ordinary t-statistic (a1=0, a2=1) and the signal log-ratio (a1=1, a2=0). The optimality is defined in terms of maximal overlap of top-ranked genes in group-preserving bootstrap datasets. Importantly, besides the group labels, no a priori information about the properties of the data is required and no fixed cutoff for the gene rankings needs to be specified. For more details about the reproducibility-optimization procedure, see Elo et al. (2008). The user is given the option to adjust the largest top list size considered in the reproducibility calculations, since lowering this size can markedly reduce the computation time. In large data matrices with thousands of rows, we generally recommend using a size of several thousands. In smaller data matrices, and especially if there are many rows with only a few non-missing entries, the size of K should be decreased accordingly. ROTS tolerates a moderate number of missing values in the data matrix by effectively ignoring their contribution during the operation of the procedure. However, each row of the data matrix must contain at least two values in both groups. The rows containing only a few non-missing values should be removed; or alternatively, the missing data entries can be imputed using, e.g., the K-nearest neighbour imputation, which is implemented in the Bioconductor package impute. If the parameter values a1 and a2 are set by the user, then no optimization is performed but the statistic and FDR-values are calculated for the given parameters. The false discovery rate (FDR) for the optimized test statistic is calculated by permuting the sample labels. The results for all the genes can be obtained by setting the FDR cutoff to 1.

References

L. L. Elo, S. Filen, R. Lahesmaa and T. Aittokallio: Reproducibility-optimized test statistic for ranking genes in microarray studies. IEEE/ACM Transactions on Computational Biology and Bioinformatics 5: 423--431, 2008.

Examples

Run this code

## ROTS-statistic for the Affymetrix spike-in data. 
rots.out <- ROTS(data = affySpikeIn, groups = c(rep(0,5), rep(1,5)),
    B = 100, K = 500 , seed = 1234)
## Summary of the ROTS results.
rots.summary <- summary(rots.out, fdr = 0.05)

Run the code above in your browser using DataLab