miRNApath (version 1.32.0)

runEnrichment: Perform gene set enrichment analysis on a miRNApath object

Description

This method performs a hypergeometric enrichment analysis on a miRNApath object, which should already contain miRNA data, miRNA-gene associations, gene-pathway associations, and some criteria for filtering miRNA hits from the full tested set.

Usage

runEnrichment(mirnaobj, Composite=TRUE, groups=NULL,
permutations=0)

Arguments

mirnaobj

An object of type mirnapath containing data resulting from the loadmirnapath method.

Composite

Defines whether the enrichment treats miRNA-gene as the enriched entity, or uses only the gene.

groups

List of groups to include in the analysis, although each group is analyzed independent from the other groups.

permutations

The number of permutations to use in calculating an adusted P-value. Value of 0 will perform no permutations.

Value

The method returns an object of type mirnapath, a list with components:

mirnaTable

data.frame containing the miRNA results data

columns

list containing the names of required column headers associated to the actual column header supplied in the dataset contained in mirnaTable. Required headers: mirnacol, assayidcol, groupcol, filterflagcol.

groupcount

The number of groups contained in mirnaTable using the groupcol, if supplied

state

The current state of the object, in this case "enriched".

mirnaGene

data.frame containing associations between miRNAs and genes.

mirnaPathways

data.frame containing gene-pathway associations.

pathwaycount

Numerical value indicating how many pathways are available in the data, provided for convenience.

filters

List of filters applied to the data, which may include: "P-value", "Fold change", and/or "Expression".

enrichment

Enrichment summary data in the form of a list of elements for each sample group (the sample group is the name of each element.) Each list element is itself a list with enrichment result data for each sample group, as independently calculated: "pvalues" - list of P-values named by pathway ID. "Measured pathway mirnaGenes" - total number of miRNA-gene-pathway combinations measured, which gives some idea of the overall coverage of pathways. The general point is that miRNAs have the potential to cover many genes and pathways. "Total mirnaGenes" - number of miRNA-gene combinations represented in the data. "Enriched pathway mirnaGenes" - number of miRNA-gene values enriched in the pathway tested. "Enriched by miRNA" - list of miRNAs involved in the pathway tested, with the list of genes in parentheses per miRNA. "Enriched by Gene" - same as previous except switching gene and miRNA. "Total enriched mirnaGenes" - the total number of miRNA-gene values involved in any pathway enrichment (significant or not.) The total values are useful when comparing across sample groups, looking particularly for groups with few changes or those with a uniquely high number of changes. Lastly, with permutations > 0 "Permutation P-value" will contain the rank-adjusted P-value as described in the details section.

pathwayList

Named list of pathways contained in the mirnaobj\$mirnaPathways object, named by the pathway ID values found in the pathwayidcol column. This list facilitates converting the data in the enrichment element to pathway names, since those values are named by the pathway ID to conserve memory.

Details

The composite flag indicates whether to treat the fully expanded miRNA-gene combinations as separate enrichment events (TRUE), or whether to treat all effects on one gene as one collective event. The latter case reverts to the classic un-ordered hypergeometric enrichment technique.

However the expansion of combinations is the current method chosen to represent the multiple predicted effects of miRNAs to one gene, and the predicted effect of one miRNA to multiple genes. The algorithm will identify statistically significantly enriched results when the combination of these effects is greater than would be anticipated by random chance.

The adjusted P-value is calculated using the rank of unadjusted P-values divided by the number of permutations minus one (such that the best rank from 1,000 permutations yields an adjusted P-value of 0.001.) The default value 0 was put in place to save time, since most adjustments resulted in stronger "hits" and weaker "non-hits" in terms of pathways enriched. Thus the results are not substantially changed, and permutation adjustment is saved for the final result set.

References

John Cogswell (2008) Identification of miRNA changes in Alzheimer's disease brain and CSF yields putative biomarkers and insights into disease pathways, Journal of Alzheimer's Disease 14, 27-41.

See Also

loadmirnapath, filtermirnapath, loadmirnatogene, loadmirnapathways

Examples

## Not run: 
# ## Start with miRNA data from this package
# data(mirnaobj);
# 
# ## Now run enrichment test
# mirnaobj <- runEnrichment( mirnaobj=mirnaobj, Composite=TRUE,
#    groups=NULL, permutations=0 );
# ## End(Not run)