Richard Holland wrote:
First thing to try to do is to get it to update/synchronise the
partition tables before generating the SQL, particularly if you have
updated your schema connection settings.
Everything is up-to-date as far as I can see. I used "mart->Update all
schemas" and "mart->Update all partition counts". The partition list is
still limited to one (the schema level partition derived from the
regexp i.e. pyrococcus) but I'm not sure if this should have changed or
not.
One point that makes me a little worried is that in the partition
dialogue, I don't get any values in the preview box, though the values
of species_id appear to make it into the final dataset names in the SQL
e.g. multi_bacterial_mart_51.pyrococcus_gene_ensembl_77__gene__main.
It'd be useful if you could
tell us how the xrefs have been partitioned, as that probably holds a
clue.
These xrefs (attached to gene as GO is involved here) appear to be
partitioned by name (I haven't altered them beyond whats already in the
ens50 xml file):
<datasetPartitionTable name="object_xref"
selectedColumns="db_name_1018">
<partitionApplication name="gene_ensembl"
dimension="gene__object_xref" pCols="db_name_1018"
dsCols="1.18.0.db_name" nameCols="\
db_name_1018" compounds="0"/>
</datasetPartitionTable>
There do appear to be GO xrefs for each of the species, so that wouldn't
seem to present a problem.
As for the subdivision boundary, if it's the first or last entry in
the list, it'll get dropped automatically as it serves no purpose in
that position. Also the idea of the boundary is to replicate groups of
columns within one table, so it isn't going to be needed here.
Ah, thanks - that clarifies things. I was mainly trying likely-looking
things that I might have missed...
Thanks again,
Dan.
--
Dan Staines, PhD
EMBL-European Bioinformatics Institute Tel: +44-(0)1223-492507
Wellcome Trust Genome Campus, Hinxton Fax: +44-(0)1223-494468
Cambridge CB10 1SD, UK email: [EMAIL PROTECTED]
