[BioSQL-l] gff in biosql; biosql, gbrowse observations

Genevieve DeClerck gad14 at cornell.edu
Thu Mar 3 16:25:48 EST 2005


Hi,

I've been playing around with a test biosql database I have running 
(with gbrowse frontend and mysql rdbms). Using the 'load_seqdatabase.pl' 
script I can successfully load a genbank file (NC_004578.gbk, the 
TIGR-annotated P. syringae chromosome) into a mysql biosql database. 
However, when I try to load a gff file with additional syringae data I 
run in to problems - it appears that GFF is not supported by biosql?? (I 
don't see it as a valid format in SeqIO/ or any mention in the docs).

Also, when I take a look at the data in the biosql database, I see a lot 
of empty tables and empty fields. The bulk of the data loads into these 
tables: location, seqfeature, seqfeature_qualifier_value. One table I 
expected to see data in is dbxref - I was surprized to see this 
completely empty because there are db_xrefs defined throughout the gbk 
file I loaded. Other tables that are empty: bioentry_dbxref, 
bioentry_path, bioentry_relationship, dbxref_qualifier_value, 
location_qualifier_value, seqfeature_dbxref, seqfeature_path  
seqfeature_relationship, term_* (all term tables except 'term').

As for empty fields, I would not have expected the 'display_name' in 
table 'seqfeature' to be empty/NULL. This appears to be an important 
field for feature display in gbrowse. When I view this biosql database 
in gbrowse I can see CDS and genes glyphs, but the identifier labels are 
missing (even thought they're acitivate in the .conf file). I had a 
suspicion that the NULL display_name field might have something to do 
with the missing glyph label so I edited a record in 'seqfeature' in 
mysql, replacing NULL with "Foo" -- I refreshed the gbrowse page and, 
sure enough, a "Foo" label appeared on the glyph whose info I altered in 
the db. This proved to me that there definately seems to be some 
disconnect between how the gbk file was parsed by load_seqdatabase.pl 
and/or how these data were inserted into the database.

Any thoughts or suggestions?

Thanks,
Genevieve






More information about the BioSQL-l mailing list