[Bioperl-l] Fwd: problem with bioperl (where's the Mus?)

[Chris Fields]

On Aug 24, 2009, at 3:01 PM, Sendu Bala wrote:

> Chris Fields wrote:
>> On Aug 24, 2009, at 2:12 PM, Sendu Bala wrote:
>>> Hilmar Lapp wrote:
>>>> On Aug 24, 2009, at 1:16 PM, Sendu Bala wrote:
>>>>>> This points to a problem in Bio::Species::scientific_name(),  
>>>>>> given that binomial() is correct. Could you file this as a bug  
>>>>>> report?
>>>>>
>>>>> What code creates the Bio::Species object here? I suspect this  
>>>>> code isn't aware of changes in Bio::Species since BioPerl 1.5.2.
>>>> I see. Any pointer to what would tell me what I need to change or  
>>>> is everything in the Bio::Species POD?
>>>
>>> ... I won't guarantee the perfection of the POD ;)
>>>
>>>
>>>> BTW what the Bioperl-db code does is instantiate the blank object  
>>>> and then populate it through its accessors (mostly the  
>>>> classification() array). If what it has been doing in the past is  
>>>> now considered incorrect, at least it doesn't raise any warning  
>>>> that would alert one to that ...
>>>
>>> Yuh... If you point out the code that creates the Bio::Species I  
>>> can look into it for you and suggest what needs changing and why  
>>> it doesn't work (or if it's a bug in Bio::Species). I can't  
>>> remember things clearly right now, though classification() I guess  
>>> was supposed to be backwards compatible.
>> Sendu, I think it's related to this:
>> http://bugzilla.open-bio.org/show_bug.cgi?id=2092#c4
>> Bio::DB::BioSQL::SpeciesAdaptor and  
>> Bio::DB::BioSQL::mysql::SpeciesAdaptorDriver are the two modules in  
>> question i think.
>
> Ah, yes, well there you go then. So it is a classification() issue.  
> Judging by what I said in that bug, looks like the db code needs to  
> be changed to put the full scientific name in the first element it  
> passes to classification.


Yup.  I believe the only blocking issue with implementing it was  
potential backwards-compat problems with databases loaded using old  
behavior and then being updated post-1.5.2 (new behavior).  I would  
think this only affects sequence data loaded w/o taxonomy preloaded,  
but I'm not sure.

I suggest, if you can fix it, go ahead make the necessary change.  We  
can then post a big warning to BioSQL and here about the problem,  
something along the lines of 'bioperl-db in svn may be backwards  
incompatible with species information loaded in previous versions; it  
may eat your first born' or similar.  It's an absolutely necessary  
fix, and may effectively kill a bunch of other db/species-related bugs.

chris