VIPER

VIPER (Virtual Inference of Protein-activity by Enriched Regulon analysis)

A multiple sample version (msVIPER) designed for gene expression signatures based in multiple samples or expression profiles, and the single sample version (VIPER) ,which estimates relative protein activity on a sample-by-sample basis, thus allowing transformation of a typical gene expression matrix.

5. Generating the regulon object

msVIPER and VIPER require a gene expression signature and an appropriate cell context-specific regulatory network.

The regulatory network is provided as the regulon object.

Regulon objects is generated by the function aracne2regulon, which takes two arguments as input: the ARACNe output .adj file and the expression data-set used by ARACNe to reverse engineer the network.

6. Master Regulator Analysis performed by msVIPER

The viper package includes the function rowTtest that efficiently performs Student's t-test for each row of a dataset. The rowTtest function conviently take an 'ExpressionSet' object as argument and produce a list object containing the Student's t-statistic and the test's p-value (p.value), that by default is estimated by a 2-tail test.

It can also take two matrixes as arguments, the first one contianing the 'test' sample and the second the 'reference' sample.

While we could define the Gene Expession signature (GES) by using the t-statistic, to be consistent with the Z-score based null model for msVIPER, we will estimate z-score values for the GES.

6.3 msVIPER

The msVIPER analysis is performed by the msVIPER function. It requires a GES, regulon object and null model as arguments, and produces an object of class 'msVIPER', containing the GES, regulon and estimated enrichment, including the Normalized Enrichment Score (NES) and p-value, as output.

Reference:
Using viper, a package for virtual inference of protein-activity by enriched regulon analysis.