metaResults: MetaRank Summary of Differential Expression Results

View source: R/metaResults.R

metaResultsR Documentation

MetaRank Summary of Differential Expression Results

Description

Combine DE results from several DuffyTools into a meta ranking of most up-regulated genes.

Usage

metaResults(targetGroup, results.path = "results", speciesID = getCurrentSpecies(),
	geneColumn = "GENE_ID", subfolderName = "Group", 
	tools = c("RoundRobin", "RankProduct", "SAM", "EdgeR", "DESeq"), 
	altGeneMapLabel = NULL,
	rank.average.FUN = sqrtmean, value.average.FUN = mean, 
	keepIntergenics = FALSE, 
	topFolders = NULL, other.DE.files = NULL,
	missingGenes=c("drop","fill","na"), nFDRsimulations=0,
	direction = c("UP","DOWN"))

metaResultsToHTML( m, fileout, title = "", maxRows = 100,
		linkColumnNames = NULL, ...)

Arguments

targetGroup

character string of the group of samples being tested for up-regulation. See details.

results.path

folder that contains all the subfolders of DE files, as made by the various DE tools.

speciesID

SpeciesID of the one species to be considered.

geneColumn

name of the column that contains the GeneIDs.

subfolderName

name of the subfolder for this set of analyses. This folder must contain all the DE results from each of the various DE tools.

tools

character vector of the DE tools that were used, to be combined into meta results.

rank.average.FUN

function to use for combining the gene rankings from all DE tools into a meta rank.

value.average.FUN

function to use for combining the gene expression values from all DE tools into an average value.

keepIntergenics

logical, use only the true genes or keep all.

topFolders

a character vector (with names) of the various folders to be combined. It is expected that each of these locations contains a results.path folder. NULL means that results.path is relative to the current working directory.

other.DE.files

an optional character vector (with names) of full pathnames to additional DE files to include in the meta rankings.

missingGenes

how to deal with genes not in all files. See metaRanks.

nFDRsimulations

how many random simulations to perform for estimating the false discovery rate (FDR).

m

data frame result from metaResults functions, to be rendered as an HTML file

fileout

Full path name for the newly created HTML file.

title

Character string on length 1, to be used as header and title in the HTML file

maxRows

Maximum number of rows to include in the HTML file

linkColumnNames

Character vector of column names, that will be converted to hyperlinks. See table2html

...

Other arguments, passed to table2html

Details

This tool combines multiple differential expression results from similar or identical datasets that were evaluated by different DE tools, to generate a consensus or meta result. The various DE tools, such as RoundRobin, RankProduct, and SAM, all use different methods and can yield somewhat inconsistent results; and this tool combines them all into one consensus meta result.

The targetGroup is one of the GroupIDs from the specified grouping column of the annotation file for that analysis. Since each DE tool uses a common naming convention for its results, all need files are locatable by the triplet subfolderName, speciesID, targetGroup.

Value

a data frame with columns for GeneIDs, products, and:

LOG2FOLD

the average fold change, as the arithmatic mean over all input DE files

AVG_PVALUE

the average P-value, as the geometric mean over all input DE files

AVG_RANK

the average rank for each gene, from all the input files, using average.FUN for calculating the average

Also, one column for each input DE file, showing the rank of each gene in that file, with column names taken from the DE tools.

If FDR simulation was done, then 2 more columns:

E_VALUE

the expected number of genes having an average rank as high as the given ranks. Expressed on the range 0 to N, where N is the number of genes in the results.

FP_RATE

the rate of false positives, as estimated by random permutations of the given ranks. Expressed on the range 0 to 1, tells the likelihood that a gene could have an AVG_RANK that good purely by chance

Author(s)

Bob Morrison

See Also

RankProduct, SAM, RoundRobin, for the various DE tools.

metaResultsToHTML, for direct creation of the HTML form of the the resulting data frame

metaRanks, for averaging of gene rank positions.


robertdouglasmorrison/DuffyTools documentation built on April 16, 2024, 6:31 a.m.