[BioC] problem with GO annotation data

[Nianhua Li]

Hi, Shiliang,

You were right that AnnBuilder did have a bug. Thanks for the report!

The bug is related to Weijun's question: 
http://thread.gmane.org/gmane.science.biology.informatics.conductor/9227/focus=9232
When baseType is "ll", the baseFile can't be parsed correctly because 
some parsers assume Entrez Gene is at the 3rd column of the base file. 
Thanks for John's patch, AnnBuilder just quietly converts the base file 
into three columns when baseType is "ll". So, weijun's problem got 
solved. Unfortunately, the parser used for "go2probe" mapping assumes a 
two-column setting for baseType "ll", therefore can't work correctly 
with the patch.

Now the problem has been solved. You can either get the up-to-date 
AnnBuilder (v 1.11.6) from svn or wait still Saturday noon to download 
it using biocLite or from website. You can also modify your local source 
AnnBuilder/R/ABPkgBuilder.R if you are in a hurry:
===================================================================
--- ABPkgBuilder.R      (v1.11.5 or before)
+++ ABPkgBuilder.R      (v1.11.6)

# Map GO ids to probe ids that are directly associated with the GO ids
 mapGO2Probe <- function(eg, baseMapType){
-     if(baseMapType == "ll"){
-            parser(eg) <- file.path(.path.package("AnnBuilder"),
-                                    "scripts", "GO2ProbeParser4LL")
-        }else{
-            parser(eg) <- file.path(.path.package("AnnBuilder"),
+        parser(eg) <- file.path(.path.package("AnnBuilder"),
                                     "scripts", "GO2ProbeParser")
-        }
         options(show.error.messages = FALSE)
         go2Probe <- try(parseData(eg, eg at go, ncol = 3))
         options(show.error.messages = TRUE)
====================================================================

Also the probe number mismatch is because your base file has duplicated 
probeset identifiers.

help it works this time :)

nianhua