Well what I am thinking is that any adaptor that sequentially reads in the whole file for each request is not a very good solution, the file adaptor essentially does this except it gives you the option to run through the file once first and save the contents in memory, meaning you're still doing a big iteration but not file I/O. Really the biggest improvement would be to add direct random-access indices/caches for the important DAS query arguments (i.e. segment range, feature ID). That applies to all file formats, but I imagine GFF-specific stuff would make it a bit easier/possible to actually construct.
TBH the GFF files I have created DAS sources have always been write-once, so I have just loaded them into a database with a script. Cheers, Andy On 7 Jan 2013, at 15:06, Dan Bolser <[email protected]> wrote: > I was thinking about the ID=, Name=, and Alias= tags in column 9. > However, to be honest I'm not sure if/how you'd want to plumb those > tags into DAS queries/responses! > > Just getting ahead of myself really! > > > Cheers, > Dan. > > On 7 January 2013 13:51, Andy Jenkinson <[email protected]> wrote: >> Hi Dan, >> >> I doubt the old CVS repo has a gff3 dedicated adapter. >> >> Not too familiar with BioPerl's GFF3 support, what does it give you, besides >> a sequential parser? What you need to be able to do is index by location for >> fast random access, either with an index file or in memory. The parent/part >> stuff does add a layer of complexity. I had plans to write one some time ago >> but didn't work on it much. >> >> Cheers, >> Andy >> >> On 7 Jan 2013, at 13:25, Dan Bolser <[email protected]> wrote: >> >>> Sorry, I omitted to mention that this is using ProServer. >>> >>> On 7 January 2013 13:25, Dan Bolser <[email protected]> wrote: >>>> Hi, >>>> >>>> Has anyone written a dedicated GFF3 adaptor? Getting GFF3 via the file >>>> adaptor seems a bit of a hack, especially as BioPerl has good GFF3 >>>> support. >>>> >>>> I read that "Customised extensions are also still available in the old >>>> CVS repository." Could that be where the GFF3 adaptor is maintained? >>>> >>>> Would be good to add this to the SVN if anyone has one :-) >>>> >>>> >>>> Cheers, >>>> Dan. >>> >>> _______________________________________________ >>> proserver-users mailing list >>> [email protected] >>> http://lists.sanger.ac.uk/mailman/listinfo/proserver-users >> _______________________________________________ DAS mailing list [email protected] http://lists.open-bio.org/mailman/listinfo/das
