Feature annotation below track

[erica-pathos]

Added an option to place label below track.

[veghp]

Thank you for this contribution. I agree it would be useful (cf. https://github.com/Edinburgh-Genome-Foundry/DnaFeaturesViewer/issues/47), for example to show features relevant for the negative strand. I'll have to review this in detail as it's a very important package for us, which will take some time, but please keep this PR open.

[Zulko]

Note that if you want all features to appear below the track, the simplest is to use ax.invert_yaxis(). This could be added to the examples.

from dna_features_viewer import GraphicFeature, GraphicRecord

features=[
    GraphicFeature(start=0, end=20, strand=+1, label="Small feature"),
    GraphicFeature(start=20, end=500, strand=+1, color="#ffcccc",
                   label="Gene 1 with a very long name"),
]
record = GraphicRecord(sequence_length=600, features=features)
ax, _ = record.plot(figure_width=4)

# Put the labels below the tracks
ax.invert_yaxis()
ax.xaxis.set_ticks_position('top')

If the goal is to have labels both above and below the track, then we must be careful with compute_feature_levels(), which should probably be called separately to compute the levels of the features above the plot and those below the plot.

[veghp]

Thank you for this, yes all features wanted below the track, at least for this issue. I'll add this as an example, perhaps mention in the main readme as well.

[erica-pathos]

Thanks for the response, the reason why I would want the label to be on the bottom (also the axis, tick labels etc), is that I want to be able to add other features on top (like lollipop plots etc.)

I guess I could add lollipop bars by adding a negative to the y axis numbers, but it would be nice to have that flexible option. and more intuitive in adding other features?

[Zulko]

Hmm yes that's a good use case, but in that case the parameter should be something like graphic_record.plot(features_below_track=True) to apply to all features at once, rather than at the feature level in GraphicFeature() which will be problematic if compute_feature_levels() isn't called twice on the two groups of features.

[erica-pathos]

I guess that works too? For my use case, the lollipops were added to the ax using generic plotting functions, in addition to the genome features, so I guess your way should work too.

Then I guess in other cases to plot two group of labels, one on top and other on the bottom, we can just do graphic_record.plot(features_below_track=True) and graphic_record.plot(features_below_track=False) on the same ax, and we can have both group labels on both side? Instead of defining each GraphicFeature to be below or above track?

Btw, I would still prefer that the features_below_track=True doesn't just invert the y axis though, when I add additional features like lollipop, the exact Y axis and the values do matter, I would like to have the option to still add those numbers as positive values.

[Zulko]

Yes to clarify what I meant was features_below_track=True that causes the plotter to transform y to -y before plotting the features, and hopefully this is enough to get them plotted in "negative" while keeping the general layout, and without needing to invert the axis. I won't have time to make an MR but that would be my preference for implementation.