[Bioperl-l] Reliability of get_Seq_by_gi

[Wes Barris]

Rob Edwards wrote:

> Did you capture the errors and do they make any sense?

I did not capture any errors.  Usually, the message says that
the GI number does not exist although I know that it does.
I think that there are general reliability problems when getting
sequences from NCBI as there are when using netblast (blastcl3).

> 
> Rob
> 
> 
> On Apr 5, 2004, at 5:44 PM, Wes Barris wrote:
> 
>> Rob Edwards wrote:
>>
>>> Just a suggestion, but have you tried making sure that $seq exists 
>>> before writing. Something like this should fix the problem:
>>>
> 
> <snip>
> 
>>> You should capture all the failed sequences and then validate them in 
>>> case there is some error with the Acc/GI (like they doesn't exist any 
>>> more).
>>> Rob
>>
>>
>> Hi Rob,
>>
>> Your solution appears to have worked.  I was able to download sequences
>> over night and my script was still running this morning -- the first time
>> it has ever done that.  Thanks.
> 
> 


-- 
Wes Barris
E-Mail: Wes.Barris at csiro.au