Right, there are two things that need to be described: the annotation track and the reference genome. The annotation track, if given a provider, version and name, always (well at least in my experience) implies a particular genome provider/version. So: TxDb.Hsapiens.myMart.64.ccds Might work? 2011/11/4 Hervé Pagès > Hi Michael, > > > On 11-11-03 06:36 PM, Michael Lawrence wrote: > >> We're actually using a patched version of makeTranscriptDbFromBiomart to >> get models out of an internal biomart. Patch is on its way to Marc. >> >> So it would be like: TxDb.Hsapiens.BioMart.hg19.**gneGenes? >> > > This suggests that you have a Mart called "hg19" (see below why). > > > >> Seems weird to mix the technical mode of data retrieval into the name. >> > > The naming scheme when 'Data source' is "BioMart" seems to be a little > bit different. For example, if I use makeTranscriptDbFromBiomart() with > biomart="ensembl" and dataset="hsapiens_gene_**ensembl", then I get: > > > GenomicFeatures:::.**makePackageName(txdb) > [1] "TxDb.Hsapiens.BioMart.**ensembl.GRCh37.p5" > > Token #4 ("ensembl") is the name of the Mart. I'm a little bit > surprised with token #5 though. I would have expected it to be > the ensembl version (eventually followed by the reference genome) > because one can always infer the reference genome from the ensembl > version but not the other way around. In other words, if Ensembl > makes 2 or more releases based on the same reference genome, our > current naming scheme won't differentiate the 2 TxDb packages. > Wouldn't it be better if we had something like: > > TxDb.Hsapiens.BioMart.ensembl.**63 > TxDb.Hsapiens.BioMart.ensembl.**64 > > Anyway, back to your problem. Yes in your case the technical mode > doesn't really matter so it's really up to you. Maybe being explicit > about the reference genome (with *.UCSC.hg19.*) is more important > than the technical mode? > > H. > > >> Michael >> >> 2011/11/3 Hervé Pagès > >> >> >> Hi Michael, >> >> >> On 11-11-02 08:58 PM, Michael Lawrence wrote: >> >> What are the precise meanings of the tokens in the TxDb package >> names. In >> particular, is "UCSC" the genome provider or the annotation >> provider? In >> the official packages, those are one in the same, but if someone >> wanted to >> make a package for custom annotations on a UCSC genome? >> >> >> The pkg name is generated automatically by internal helper function >> GenomicFeatures:::.__**makePackageName(). This function extracts all >> the >> >> tokens from the txdb's metadata table. It looks like the 3rd token >> in the pkg name is extracted from the 'Data source' field and can only >> be "UCSC" or "BioMart", typically indicating whether the txdb was made >> with makeTranscriptDbFromUCSC() or makeTranscriptDbFromBiomart(). >> The first function downloads annotations from the UCSC genome >> browser using rtracklayer. The 2nd one downloads them with biomaRt >> from whatever mart/dataset was specified. >> >> For your custom annotations, the final name of the pkg will depend on >> what GenomicFeatures:::.__**makePackageName() finds in the metadata >> >> table of your txdb, but, if 'Data source' is not "UCSC" or "BioMart", >> it seems that GenomicFeatures:::.__**makePackageName() will fail (and >> not >> >> in a very informative way I'm afraid). If I understand correctly, you >> are making your custom txdb object with a call to makeTranscriptDb()? >> If that's the case, make sure you provide enough information >> thru its 'metadata' argument. Maybe you could set 'Data source' to >> "UCSC" and use some kind of custom name for the table (which in your >> case is probably not a real UCSC "table"). This custom name will become >> the last token in the package name. So you would end up with something >> like: >> >> TxDb.Hsapiens.UCSC.hg19.__**GenentechGenes >> >> >> This solution would have the advantage of having >> GenomicFeatures:::.__**makePackageName() work out-of-the-box. >> >> But maybe it's confusing because it suggests that >> the txdb was made with makeTranscriptDbFromUCSC()? I hope >> it's not. >> >> H. >> >> >> Thanks, >> Michael >> >> [[alternative HTML version deleted]] >> >> ______________________________**___________________ >> Bioc-devel@r-project.org >> > >> mailing list >> https://stat.ethz.ch/mailman/_**_listinfo/bioc-devel >> >> >> > >> >> >> >> -- >> Hervé Pagès >> >> Program in Computational Biology >> Division of Public Health Sciences >> Fred Hutchinson Cancer Research Center >> 1100 Fairview Ave. N, M1-B514 >> P.O. Box 19024 >> Seattle, WA 98109-1024 >> >> E-mail: hpages@fhcrc.org >> Phone: (206) 667-5791 >> Fax: (206) 667-1319 >> >> >> > > -- > Hervé Pagès > > Program in Computational Biology > Division of Public Health Sciences > Fred Hutchinson Cancer Research Center > 1100 Fairview Ave. N, M1-B514 > P.O. Box 19024 > Seattle, WA 98109-1024 > > E-mail: hpages@fhcrc.org > Phone: (206) 667-5791 > Fax: (206) 667-1319 > [[alternative HTML version deleted]]