[BioSQL-l] Can I just consider import HGI.101403 to biosequence table?

[Hilmar Lapp]

On Wednesday, December 17, 2003, at 10:50  AM, Josh Lauricha wrote:

> The select took basically no time, but the time spent inside Bio::* was
> significant. This is on quad xeon 1.9GHz with 8GB RAM... so it was
> probably something with the local setup, or I'm confusing things.
>

Hm. Try to execute the search with a unique key query 
($adp->find_by_unique_key()). There should be no difference. If there 
is then it hints at a bioperl-db problem, or at a missing or wrong 
index on accession number.

But you say the select was fast. Note that by the time you get the 
record returned as an object, several tens of selects will have been 
executed to fetch all the related entities. The only time this would 
take several seconds is if your accession is for a unigene cluster with 
hundreds or even thousands of members, because each member is a 
bioentry itself.

Note that you can turn on a lot (!) of debugging output by setting 
$db->verbose(1) before you obtain any adaptor or persistent object from 
it. If you then watch the log online, you may see if there's any 
problem query where it stalls briefly. Some people have reported that 
the species retrieval in I believe mysql was a serious bottleneck (and 
then it is the sql query itself). Species are cached once they've been 
retrieved once, but if you restart the script every time you won't take 
advantage of this.

	-hilmar
-- 
-------------------------------------------------------------
Hilmar Lapp                            email: lapp at gnf.org
GNF, San Diego, Ca. 92121              phone: +1-858-812-1757
-------------------------------------------------------------