[BioPython] Fwd: NCBI Abuse Activity with BioPython

[Chris Fields]

On Jun 27, 2008, at 8:57 AM, Peter wrote:

> On Fri, Jun 27, 2008 at 2:19 PM, Frederico Arnoldi <fredgca at hotmail.com 
> > wrote:
>>
>> Guys (sorry the informality),
>>
>> I have followed the discussion about "NCBI Abuse Activity with  
>> BioPython". I
>> have to confess that followed it superficially, since I am not able  
>> to understand
>> everything you said. So, I am going to make some questions about it:
>>
>> 1)I believe that using BLAST with NCBIWWW.qblast is included in  
>> "Abuse Activity". Right?
>
> I'm not aware that abuse of BLAST was singled out, only Entrez / E- 
> utils.

Similar policy though, for the same reasons they insist on a delay for  
E-utils.

>> I am asking because sometimes I use it. The recommendation of NCBI is
>> "Make no more than one request every 3 seconds.".
>
> True, http://www.ncbi.nlm.nih.gov/blast/Doc/node60.html
>
>> Biopython code does not assure it with the following  code in  
>> NCBIWWW.py,
>> line 779:
>> [code]
>> limiter = RequestLimiter(3)
>> while 1:
>>   limiter.wait()
>> [/code]
>
> I believe that bit of code is polling the server for results every
> three seconds.  Perhaps we should insert an additional enforced three
> second delay between submission of queries as well.
>
>> 2)Do you have any recommendation for using it that it is not  
>> included in the
>> tutorial? Maybe listing some recommendations here would help.
>
> I would recommend running your own local BLAST server for any large
> jobs - either the standalone blast tools, or if you have a machine on
> the network that many people could share, run the WWW version locally.
>
> Peter

The above appears to submit a single job at a time and wait 3 sec.  
between polling the server until the current job is finished.  I don't  
think that is the problem indicated in the link above.  The 3 sec. is  
for submitting new BLAST jobs, for instance if you want to submit one  
BLAST request after another (gathering the RIDs), then grab all the  
reports at once, or if you are threading 50 submission requests all at  
once.

chris