gosummaries.MArrayLM: Prepare gosummaries object based on limma results
In raivokolde/GOsummaries: Word cloud summaries of GO enrichment analysis

Description Usage Arguments Details Value Author(s) Examples

The gosummaries object is created based on the differentially expresed genes, each contrast defines one component.

## S3 method for class 'MArrayLM'
gosummaries(x, p.value = 0.05, lfc = 1,
  adjust.method = "fdr", exp = NULL, annotation = NULL,
  components = 1:ncol(x), show_genes = FALSE, gconvert_target = "NAME",
  n_genes = 30, organism = "hsapiens", ...)

`x`	an object of class `MArrayLM`
`p.value`	p-value threshold as defined in topTable
`lfc`	log fold change threshold as defined in topTable
`adjust.method`	multiple testing adjustment method as defined in topTable
`exp`	an expression matrix, with row names corresponding to the names of the genes in clusters (Optional)
`annotation`	a `data.frame` describing the samples, its row names should match with column names of `exp` (Optional)
`components`	numeric vector of comparisons to annotate
`show_genes`	logical showing if GO categories or actual genes are shown in word clouds
`gconvert_target`	specifies gene ID format for genes showed in word cloud. The name of the format is passed to `gconvert`, if NULL original IDs are shown.
`n_genes`	maximum number of genes shown in a word cloud
`organism`	the organism that the gene lists correspond to. The format should be as follows: "hsapiens", "mmusculus", "scerevisiae", etc.
`...`	GO annotation filtering parameters as defined in `gosummaries.default`

The usual differential expression analysis involves making several comparisons between treatments ehere each one yields an up and down regulated gene list. In a GOsummaries figure each comparison is displayed as one component with two wordclouds. If expression matrix is attached then the panel shows the expression values for each gene as boxplots, if not then number of genes is displayed

It is possible to show the gene names instead of GO annotations in the wordclouds. The word sizes in wordclouds are defined by the limma p-values. As the gene identifiers in expression matrices are usually rather unintelligible then they are automatically converted into gene names using gconvert function. It is possible to show also the original identifiers by setting gconvert_target to NULL. This can be useful if the values do not correspond to genes, but for example metabolites.

A gosummaries object.

Raivo Kolde <raivo.kolde@eesti.ee>

## Not run: 
data(tissue_example)

# Do the t-test comparisons
mm = model.matrix(~ factor(tissue_example$annot$Tissue) - 1)
colnames(mm) = make.names(levels(factor(tissue_example$annot$Tissue)))

contrast = limma::makeContrasts(brain - cell.line, 
                                hematopoietic.system - muscle, 
                                cell.line - hematopoietic.system, 
                                levels = colnames(mm))

fit = limma::lmFit(tissue_example$exp, mm)
fit = limma::contrasts.fit(fit, contrast)
fit = limma::eBayes(fit)

gs_limma = gosummaries(fit)
gs_limma_exp = gosummaries(fit, exp = tissue_example$exp, 
                           annotation = tissue_example$annot)

plot(gs_limma, fontsize = 8)
plot(gs_limma, panel_height = 0, fontsize = 8)
plot(gs_limma_exp, classes = "Tissue", fontsize = 8)

## End(Not run)