Fields, Christopher J | 4 Mar 05:50 2013

Re: module for description of sequence variants (where to place code)

On Mar 3, 2013, at 10:15 PM, Carnë Draug <carandraug+dev <at> gmail.com>
 wrote:

> On 4 March 2013 03:45, Fields, Christopher J <cjfields <at> illinois.edu> wrote:
>> [...] at some point it might be worth thinking about whether we need to
>> organize the various *Utils modules a bit better.  As most of these export
>> methods, they would be good targets for reorganization at some point
>> (maybe into a general Bio::Utils namespace).
> 
> Do you mean placing all of them into a separate Util distribution? I
> think it makes more sense to keep them close to what they deal with.

No, I agree we keep them in core.  I meant we should think about what specific domain the various methods
contained within those modules deal with, and whether it would be best.  For instance, we also have an
AlignUtils, TreeUtils, etc; at some point it might be easier to collect any of the *exportable* methods
together in one common space.  At the moment they're a bit scattered, some are exportable, some not…

> If it's just moving them all into Bio::Utils:: may be a good idea. At
> the moment there's not much of a model for their naming. I found these
> ones (am I missing anything that does not have Utils on the name?)
> 
> Bio/Coordinate/Utils.pm
> Bio/SeqUtils.pm
> Bio/Search/SearchUtils.pm
> Bio/Search/BlastUtils.pm
> Bio/Search/Tiling/MapTileUtils.pm
> 
> Also, only the last one seems to export anything.
> 
> Carnë

A few mentioned above.  I don't think they have to be object classes, their entire purpose is to add
functionality, so a full-blown class is probably a bit overkill (no one is likely to override their
behavior).  

I could also see subdividing these up: DNA, RNA, and protein specific methods as well (for instance the 1->3
letter amino acid conversions, which would never be used on DNA).  It just depends how fine-grained we want
to get.  

chris

Gmane