[BioSQL-l] bioentry: put molecule or division back?

[Yves Bastide]

Hilmar Lapp wrote:
> If you need to tell nucleotide from protein, wouldn't the alphabet tell 
> you? I.e., either alphabet is already set at the time you get the object 
> in hand, or if not, you set it based on molecule(). Would that work or 
> not? If you have neither alphabet nor molecule, how would you know what 
> to ask Entrez for?
> 

I don't have the alphabet anymore at the time I need it, only the 
accession number.

 From a local nucleic sequence, I did a blastp or a tblastn, which gave 
back (accession number, description) with which I added a tupple into 
bioentry.  Later, the corresponding sequence and annotations may be 
requested.

Hmm, upon reflection, I see four alternatives:

1. use the blast information (sbjct->molecule() or just program name): 
ugly as it add a dependency

2. add molecule to bioentry: too extreme

3. half-fill a biosequence (alphabet and length): ugly

4. use the biodatabase

(4) seems the better option, as it should also save the embarrassment of 
  querying Entrez with a SP accession...

</hand-thinking>


>     -hilmar
> 

yves

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