Skip to content
Prev 607 / 21312 Next

[Bioc-devel] Making GOstats pathway analysis using data from both a and b chip s in one go.

Hi,
   Basically the problem with combining data from multiple chips is that 
you need to be pretty careful about duplicate probes for single genes 
(since GO maps at the EntrezGene level that is the level of resolution 
you have to work with and double counting some genes and not others is a 
pretty bad idea) and next it is getting the Universe right.  This is 
also incredibly important, and is straightforward for a single chip (and 
that is what has been done in GOstats) and less straightforward for 
anything else.  You will need to do some programming to create your own 
Universe and solve a few other problems (the GOHyper has a universe 
argument that should work, let us know if it does not).

  Finally as Wolfgang recently said, this approach (hypergeometric 
testing) is probably not as good as Gene Set Enrichment (there are two 
Bioc packages for doing this, one is called Category the other is called 
sigPathway

  best wishes
    Robert
Peder.Worning at astrazeneca.com wrote: