added test for negative color from bigwig or bedgraph tracks

[fidelram]

No description provided.

[fidelram]

@Phlya I added some tests for the new feature that you added. However, I see some problems with the bigwig fill case. See the first track in this image:

Can you take a look? I added a bigwig file that has some negative values to the test data. You can your the test as py.test test_bigWigTrack.py.

In the bedgraph case I don't see the problem, but for type=line and points the image looks strange to me:

[Phlya]

Thank you, I'll try to have a look tomorrow!

[Phlya]

Sorry, I didn't manage to look into it then, and now I'm on holidays with not much free time. But I remember about this.

[Phlya]

OK, for some reason interpolate=True creates this weird problem for bigwig tracks... Strange. I'll try some simple examples with matplotlib plots, maybe it's a bug...

[Phlya]

Yep, pretty sure this is a bug in matplotlib!

[Phlya]

So the issue is having NaNs... While it has been confirmed a bug in matplotlib, it should be possible to also create a workaround while they are sorting it out - maybe set all NaNs to zero? Or to a value of the opposite sign to what we are plotting?

[fidelram]

I think that setting the values to zero may work.

…

On Sat, Jul 28, 2018 at 10:27 AM Ilya Flyamer ***@***.***> wrote: So the issue is having NaNs... While it has been confirmed a bug in matplotlib, it should be possible to also create a workaround while they are sorting it out - maybe set all NaNs to zero? Or to a value of the opposite sign to what we are plotting? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#37 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AEu_1UI0PN85G0SKM2q8kWVZo_vwPxzGks5uLCB4gaJpZM4VG3FW> .

[fidelram]

I would like to add a new pyGenomeTracks release. Do you think we can finalize this? Using interpolation=False does not work?

[Phlya]

Hi, sorry, forgot about this to be honest. See my PR, nans get replaced by zeros. I ran the test and it looks fine.

[Phlya]

FYI, as discovered in mpl issue tracker, updating numpy to 1.16 fixes the issue with interpolation.
matplotlib/matplotlib#11781 (comment)

[fidelram]

Great to know. I will update the requirements

[Phlya]

Well, we circumvented the issue by converting all NaNs to zeros, so it is fine now anyway, but you might want to revert https://github.com/deeptools/pyGenomeTracks/pull/37/files/47b13185690f65e53f27f92666095ed3ab5d2048..HEAD then if you are updating requirements.