Hi, Weijun, Thanks for the information. It is a bug. The code uses "ftp://ftp.ncbi.nlm.nih.gov/gene/DATA/gene2accession.gz" for refseq to Entrez Gene mapping. It should use "gene2refseq.gz" instead. The bug has been fixed and version number increased to 1.11.5. The update is available on svn now, and will be available on web at around 1pm, July 26 (Pacific Time). If you are in a hurry, this is the modification: ================================================== --- ABPkgBuilder.R (v1.11.4) +++ ABPkgBuilder.R (v1.11.5) @@ -558,11 +557,12 @@ } getUniMappings <- function(baseName, eg, ug, otherSrc, baseMapType){ - base <- getBaseFile(baseName) if(baseMapType == "refseq"){ - unified <- unifyMappings(base, eg, NULL, otherSrc) + eg_refseq <- eg + eg_refseq at accession <- eg_refseq at refseq + unified <- unifyMappings(base, eg_refseq, NULL, otherSrc) }else if(baseMapType != "ll"){ #ug <- UG(srcUrl = srcUrl(ug), # parser = parser(ug), baseFile = baseName(ug), =================================================== Also, notice that the ACCNUM environment generated by AnnBuilder is a copy of your input, not probe-to-GenBank mapping. ABPkgBuilder doesn't provide probe-to-GenBank mapping unless baseFile contains that information. hope it works this time :P nianhua