-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Out of curiosity, when you say that for some main table entries there are no corresponding dimension entries, do you mean that corresponding rows simply do not exist? It's a mart requirement that dimension rows must always exist for every main row even if all the columns in the dimension row (apart from those in the foreign key to the main table) are null.
Also are you dealing with a mart with only one main table, or does it have secondary/additional/subclass tables (depending on your choice of terminology)? There are sometimes issues with the latter where two main tables have the same column names but different values in those columns, so that when a dimension off the subclass table is selected you suddenly see different results in the query. Hence we have a mart requirement to ensure that your second subclassed main table contains all the exact same columns and data from the parent table, plus some extra columns of its own, rather than replacing any data in columns from the parent table. Could you point us to a database server where we can login and investigate the table structure and contents? If the structure and contents are all fine then we will investigate the API code, but I strongly suspect this is a data-related problem. cheers, Richard David Croft wrote: > Hi, > > Somebody else using the Reactome Mart stumbled on something odd. > If I select an attribute from a dimension table, then click the "Results" > button, I get a ton of results. For some of the main table entries, > there is nothing present in the dimension table. In this case, I see > empty cells in the displayed results table. > > Now, if I chose the ID of one of these main table entries where there > is nothing in the dimension table, and use it as a filter term, then > re-run, I get an empty results table, even though the columns from > the main table would have contained something. Is this a feature, > or have I messed something up? > > You can try this at the following web site: > > http://banon.cshl.edu:5555/biomart/martview/ > > Select dataset "reaction". Select extra attribute "Molecule ChEBI ID" > in the section "Small molecule". Hit "Results". Everything is fine. > > Now choose the filter "Limit to reactions containing these IDs". Select > "Reaction stable ID(s)...". Enter "REACT_9447" as the filter value. > Hit "Results" again. You get nothing! > > Now, just for fun, go back to attribute selection, and de-select > "Molecule ChEBI ID", and then re-run. Hey presto! More results than > I've had hot dinners. > > Does anybody know what's going on? > > Cheers, > > David. > -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2.2 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFGSyvr4C5LeMEKA/QRAmszAJwNytNz0aWWMl/1k5V5N+2IqeXmPQCgo1OO AqblqctcBFtnEI17TnDx2yM= =Cf+x -----END PGP SIGNATURE-----
