gt gene set testing methods: Gene set testing of gene set databases using Global Test

Description

A collection of procedures for performing the Global Test on gene set databases. Three function are provided for KEGG, for Gene Ontology and for the Broad Institute's gene sets.

Usage

gtKEGG (response, exprs, ..., id, annotation, probe2entrez, multtest = c("Holm", "BH", "BY"), sort = TRUE)
gtGO (response, exprs, ..., id, annotation, probe2entrez, ontology = c("BP", "CC", "MF"), minsize=1, maxsize=Inf, multtest = c("Holm", "focuslevel", "BH", "BY"), focuslevel = 10, sort = TRUE)

gtConcept (response, exprs, ..., annotation, probe2entrez, conceptmatrix, concept2name = "conceptID2name.txt", entrez2concept = "entrezGeneToConceptID.txt", threshold = 1e-4, share = TRUE, multtest = c("Holm", "BH", "BY"), sort = TRUE)
gtBroad (response, exprs, ..., id, annotation, probe2entrez, collection, category = c("c1", "c2", "c3", "c4", "c5"), multtest = c("Holm", "BH", "BY"), sort = TRUE)

Arguments

response

The response variable of the regression model. This is passed on to the response argument of gt.

exprs

The expression measurements. May be ExpressionSet or matrix. Passed on to the alternative argument of gt.

...

Any other arguments are also passed on to gt.

The identifier(s) of gene sets to be tested (character vector). If omitted, tests all gene sets in the database.

annotation

The name of the probe annotation package for the microarray that was used, or the name of the genome wide annotation package for the species (e.g. org.Hs.eg.db for human). If an organism package is given, the argument probe2entrez must be supplied. If annotation is missing, the function will attempt to retrieve the annotation information from the exprs argument.

probe2entrez

Use only if no probe annotation package is available. A mapping from probe identifiers to entrez gene ids. May be an environment, named list or named vector.

multtest

The method of multiple testing correction. Choose from: Benjamini and Hochberg FDR control (BH); Benjamini and Yekutieli FDR control (BY) or Holm familywise error fontrol (Holm). For gtGO also the focus level method is available. See focusLevel.

sort

If TRUE, sorts the results to increasing p-values.

ontology

The ontology or ontologies to be used. Default is to use all three ontologies.

minsize

The minimum number of probes that may be annotated to a gene set. Gene sets with fewer annotated probes are discarded.

maxsize

The maximum number of probes that may be annotated to a gene set. Gene sets with more annotated probes are discarded.

focuslevel

The focus level to be used for the focus level method. Either a vector of gene set ids, or a numerical level. In the latter case, findFocus is called with maxsize at the specified level to find a focus level.

collection

The Broad gene set collection, created by a call to getBroadSets.

conceptmatrix

The name of the file containing the importance weights, i.e. concept profile associations between Anni concepts. In the matrix contained in the file, columns correspond to testable concepts, and rows correspond to entrez-concepts. Useable files can be downloaded from http://biosemantics.org/index.php/software/weighted-global-test.

concept2name

The name of the file containing a mapping between Anni concepts and entrez identifiers. Useable files can be downloaded from http://biosemantics.org/index.php/software/weighted-global-test.

entrez2concept

The name of the file containing a mapping between Anni concept numbers and names. Useable files can be downloaded from http://biosemantics.org/index.php/software/weighted-global-test.

threshold

The relevance threshold for importance weights. Importance weights below the threshold are treated as zero.

If TRUE, the function divides the importance weight of a gene over all probes corresponding to the same entrez identifier. If FALSE, all probes get the full importance weight of the gene.

Value

gt.object.

Details

These are utility functions to make it easier to do gene set testing of gene sets available in gene set databases. The functions automatically retrieve the gene sets, preprocess and select them, perform global test, do multiple testing correction, and sort the results on the basis of their p-values.

The four functions use different databases for testing. gtKEGG and gtGO use KEGG (http://www.genome.jp/kegg) and GO (http://www.geneontology.org); gtConcept uses the Anni database (http://www.biosemantics.org/anni), and gtBroad uses the MSigDB database (http://www.broadinstitute.org/gsea/msigdb). The gtConcept function differs from the other three in that it uses association weights between 0 and 1 for genes within sets, rather than having a hard cut-off for membership of a gene in a set.

All functions require that annotate and the appropriate annotation packages are installed. gtKEGG additionally requires the KEGG.db package; gtGO requires the GO.db package; gtBroad requires the user to download the XML file "msigdb_v2.5.xml" from \ http://www.broad.mit.edu/gsea/downloads.jsp, and to preprocess that file using the getBroadSets function. gtConcept requires files that can be downloaded from http://biosemantics.org/index.php/software/weighted-global-test.

References

Goeman, Van de Geer, De Kort and Van Houwelingen (2004). A global test for groups of genes: testing association with a clinical outcome. Bioinformatics 20 (1) 93-99.

Goeman, Oosting, Cleton-Jansen, Anninga and Van Houwelingen (2005). Testing association of a pathway with survival using gene expression data. Bioinformatics 21 (9) 1950-1957.

Examples

Run this code

    # Examples in the Vignette

Run the code above in your browser using DataLab