>What I did was to simply use the annotation available from Affymetrix to >map each probe id to its corresponding genbank id using the file: >http://www.affymetrix.com/Auth/analysis/downloads/taf/Mouse430_2_anno t_csv.zip >and then run ABPkgBuilder as detailed in the vignette "How to use >AnnBuilder." Yes, you were doing the right thing. I also extract the mappings between probe ids and LocusLink ids from the same file and use the data as "otherSrc" to obtain the unified mappings. >The reason for my concern is that the QC stats my efforts returned >indicate that I have fewer probes annotated then the sum of the QC stats >for the moe430a and moe430b packages would give me. The data package in the devel track (can be downloaded as described by James MacDonald) gives more mappings than you have mainly because unified mappings was used to build the data package. If we look at mouse4302LOCUSID, I got 37302 out of 45101 while you got 31362 out of 45102.