gsea.fix.output: Fix the output from a GSEA run

In drmjc/metaGSEA: Meta-analysis of GSEA analyses, including intra- and inter-experiment comparisons

Description

GSEA and GseaPreRanked both produce rich HTML reports with hyperlinks to annotation databaes, images and heatmaps. GseaPreRanked, misses out on some of the features of the GSEA reports, and both GSEA and GseaPreRanked output can be improved somewhat (IMHO).

Usage

gsea.fix.output(dir, genes.gct = NULL, fix.index = TRUE,
  fix.posneg = TRUE, fix.reports = TRUE, debug = FALSE,
  overwrite.heatmap.images = FALSE, verbose = FALSE)

Arguments

dir	see details
genes.gct	The path to a GCT file. see details.
fix.index	logical: fix the index.html file? see details.
fix.posneg	logical: Fix the pair of up-/down-regulated geneset summary tables? see details.
fix.reports	logical: fix the individual geneset reports? see details
debug	logical: print debugging messages?
overwrite.heatmap.images	logical: overwrite already existing heatmap images?
verbose	logical: print verbose messages?

Details

Here's a list of the fixes, all to the HTML files:
GSEA and GseaPreRanked:
- Within the GeneSet column, in the geneset reports, remove the link to Affymetrix, and add hyperlinks to Entrez and GeneCards
- Within the geneset reports, add a hyperlink to the genesets in the header section
- Within the up/down-regulated summaries, add hyperlinks to MSigDB for all genesets, not just the top N that had individual reports made
- Within the geneset reports, add a gene decscription into the appropriate column, by searching within the chip file listed in the rpt file
GseaPreRanked only:
- Within ‘index.html’, instead of naming the phenotypes na_pos and na_neg, call them up- and down-regulated genesets
- within the individual geneset reports, add a heatmap (if a GCT file is provided)

Value

none. it just fixes gsea outputs in situ

Control over which elements are fixed

fix.posneg: fix the 2 html files that summarise the stats for the up/down regulated genesets. This only makes sense for GseaPreRanked output, but no harm in leaving it TRUE for gsea output.
fix.index: fix the index.html file? This only makes sense for GseaPreRanked output, but no harm in leaving it TRUE for gsea output.
fix.reports: fix the individual geneset reports? This controls whether the geneset report fixes reported above, including adding heatmaps is done.

You can import individual, or multiple GSEA runs via the dir parameter:
- a single gsea directory, eg "./GSEA/pos6_versus_neg6/c1_all.Gsea.1252052322484"
- the parent dir containing multiple GSEA runs, eg "./GSEA/pos6_versus_neg6" which will fix all GSEA dirs within it.
- a single geneset report file, eg "./GSEA/pos6_versus_neg6/c1_all.Gsea.1252052322484/CHR1P21.html"

If you ran GseaPreRanked, then you don't get any heatmaps. Specifying a gct file via the genes.gct parameter will make heatmaps for each of the individual geneset reports. NB: since GSEA results are at the gene-set leve, the gct file must also be at the gene-symbol level, ie, it should have one row per gene symbol, the row names should be gene symbols, NOT the probesetID. This means you'll need to collapse your probe-level GCT file to gene-level. see collapse.gct. TODO: allow a probe-level gct file which matches the probe-level rnk file & which is generally easier to obtain a priori. This would be easier with a GCT, ProbeLevelGCT, GeneLevelGCT objects.

Author(s)

Mark Cowley, 2009-07-09

drmjc/metaGSEA documentation built on Aug. 8, 2020, 1:53 p.m.