[Bioperl-l] Bio::DB::SeqFeature sequences with no identifier?

[Fields, Christopher J]

On May 23, 2014, at 1:21 AM, Mark Wilkinson <markw at illuminae.com> wrote:

> On 22/05/2014 7:05 PM, Fields, Christopher J wrote:
>> The specification *allows* for FASTA to be stored with the features at the end, but it does not *require* it.  I personally never store them together, but my preferences differ from others, which is probably why this is made to be flexible.
> 
> We're actually talking about different things.  The loader script allows you to pass filenames for both/either GFF and FASTA files. If I pass *just* FASTA, I get a bunch of un-named sequences in my database.... which IMO is "just plain wrong"!  LOL!
> 
> M

S’what happens when I start reading the thread a little late. :P

Re: not using raw FASTA along, I’ll have to try that out but I agree, that is a definite bug (as you mentioned it should at least do a INSERT OR IGNORE at the sequence indexing step and create a minimal region in cases like that).  

I think if you use the 'in-memory’ database along with a FASTA, it will index the FASTA (using Bio::DB::Fasta) and the seqs are available.  Of course this also indicates the intent is to make seq data somehow available even in the absence of features.

chris