[Biopython-dev] Genome Diagram Default Behavior

[Peter Cock]

On Wed, Aug 1, 2012 at 3:27 PM, Zachary Charlop-Powers
<zcharlop at mail.rockefeller.edu> wrote:
> Leighton,
> Peter,
>
> I love that we're not in the same timezone; I ask a question when I leave
> work and - lo,and, behold - when I return in the morning there is a well
> thought out response.  Thank you both.

:)

Peter wrote:
>>> The good news is the underlying drawing code can do this - the
>>> arrow drawing is just given a bounding box and the requested
>>> orientation (left or right) argument set by the get_feature_sigil
>>> method of the LinearDrawer or CircularDrawer.
>>>
>>> If you need this right now, a careful hack in get_feature_sigil is
>>> the way to proceed.

Zachary wrote:
> I will take a look at this for a quick hack for some drawing I am
> working on.

I hope you found any effort spent useful for understanding the
codebase... even if it doesn't turn out to be needed (see below).

Peter wrote:
>>> That was another idea I was considering. Under this model, the sigils
>>> could be given the full strand straddling bounding box, and decide if
>>> they will use all of this (i.e. the new 'large_arrow', or the current sigils
>>> when strand-less), or just half as in the stranded current 'arrow' and
>>> 'box' sigils where the strand is known.
>>>
>>> That could work quite well, and the end user API is quite clean.

Leighton wrote:
>> This option gets my vote.
>>
>> L.

Zachary wrote:
> If you are both in agreement that this option is desirable and that it can
> be implemented in the sigil style, now we face the question of coding it.
> Would either of you consider working on it? If not this might be a problem I
> could tackle with a small amount of mentoring. Please let me know - I am
> happy to take a stab at it.

I had a go this afternoon (a quite moment between rushes - grin), and
it wasn't as bad as I feared. This is on a git branch at the moment,

https://github.com/peterjc/biopython/tree/gd-big

Thus far, just two commits. The first refactors the current code to move
the strand handling into the sigil code (but should, I hope, have no
side effects):

https://github.com/peterjc/biopython/commit/d9c416be7dd2c7081bd66bd553c9feb0174ecc13

The second commit implements the new axis straddling arrow (for
both linear and circular diagrams) plus a minimal test:

https://github.com/peterjc/biopython/commit/b58903d5c455416028a8ae410b2063d536448d59

To match the current sigil argument names BOX and ARROW, I have
provisionally called BIGARROW. Any better ideas?

Also, to match the current arrow's behaviour, strand-less features
get an arrow pointing to the right (like a forward strand arrow).
Leighton and I had a little debate about this - with hindsight, the
original arrow sigil might have raised an error or drawn a box in
this situation - but I'm not willing to change this and break existing
code.

It would be great if you (Zachary) could give this a test, both to look
for regressions (anything that broke) and try the new sigil out. Are
you familiar with git, and installing Biopython from source?

Regards,

Peter