[emboss-dev] GFF3 in EMBOSS

[Peter Rice]

Hi Pjotr,

On 12/08/10 11:33, Pjotr Prins wrote:
> I am having a look at the GFF3 implementation in EMBOSS - mostly
> ajax/core/ajfeat.c.
>
> All features are loaded into RAM, and also the sequence information,
> when in the file. Not only for GFF3, but for all feature data types.
>
> On regular desktops this is a problem when loading a larger set,
> and/or multiple genomes.
>
> Is it the idea to load big data and store it in a SQL database? I.e.
> should I recommend handling it outside EMBOSS?

We are looking into storing data structures for large datasets on disk - 
not only for features but also for next-generation mapped reads.

Can you give an example of the input you are trying to handle?

I hope to explore these issues at the GMOD meeting in Cambridge (UK) soon.

regards,

Peter Rice