Fix request: incorrect arrow inference when zoom in to a small region of a large genome

[Hocnonsense]

Hi, thanks for this usefull tool, it's useful for me.

However, I want to use it to plot a small region (~1k bp) of the genome (1M bp), and I've found there is no arrow on the genome. It is caused by the misuseage of self.sequence_length in L160:

DnaFeaturesViewer/dna_features_viewer/GraphicRecord/MatplotlibPlottableMixin.py

Lines 156 to 162 in 990b201

	if is_undirected or head_is_cut:
	head_length = 0.001
	else:
	width_pixel = self._get_ax_width(ax, unit="pixel")
	head_length = 0.5 * width_pixel * feature.length / self.sequence_length
	head_length = min(head_length, 0.6 * feature.thickness)

Example

1. a small genome

record = GraphicRecord(
    sequence=("N" * 2000000)[1999000-1000:2000000],
    features=[
        GraphicFeature(start=1999000-1999000 + 1000, end=1999500-1999000 + 1000, strand=+1, color="#ffcccc"),
        GraphicFeature(start=1999500-1999000 + 1000, end=2000000-1999000 + 1000, strand=+1, color="#ccccff"),
    ],
)

ax, _ = record.plot(figure_width=6, x_lim=(1000, None))
assert isinstance(ax.figure, plt.Figure)
ax.figure.savefig("sequence_and_translation.png", bbox_inches="tight")

record = GraphicRecord(
    sequence=("N" * 2000000)[1999000-10000:2000000],
    features=[
        GraphicFeature(start=1999000-1999000 + 10000, end=1999500-1999000 + 10000, strand=+1, color="#ffcccc"),
        GraphicFeature(start=1999500-1999000 + 10000, end=2000000-1999000 + 10000, strand=+1, color="#ccccff"),
    ],
)

ax, _ = record.plot(figure_width=6, x_lim=(10000, None))
assert isinstance(ax.figure, plt.Figure)
ax.figure.savefig("sequence_and_translation.png", bbox_inches="tight")

2. a bigger one

record = GraphicRecord(
    sequence=("N" * 2000000)[1999000-10000:2000000],
    features=[
        GraphicFeature(start=1999000-1999000 + 10000, end=1999500-1999000 + 10000, strand=+1, color="#ffcccc"),
        GraphicFeature(start=1999500-1999000 + 10000, end=2000000-1999000 + 10000, strand=+1, color="#ccccff"),
    ],
)

ax, _ = record.plot(figure_width=6, x_lim=(10000, None))
assert isinstance(ax.figure, plt.Figure)
ax.figure.savefig("sequence_and_translation.png", bbox_inches="tight")

3. a large one

record = GraphicRecord(
    sequence=("N" * 2000000)[1999000-100000:2000000],
    features=[
        GraphicFeature(start=1999000-1999000 + 100000, end=1999500-1999000 + 100000, strand=+1, color="#ffcccc"),
        GraphicFeature(start=1999500-1999000 + 100000, end=2000000-1999000 + 100000, strand=+1, color="#ccccff"),
    ],
)

ax, _ = record.plot(figure_width=6, x_lim=(100000, None))
assert isinstance(ax.figure, plt.Figure)
ax.figure.savefig("sequence_and_translation.png", bbox_inches="tight")

record = GraphicRecord(
    sequence=("N" * 2000000)[:2000000],
    features=[
        GraphicFeature(start=1999000, end=1999500, strand=+1, color="#ffcccc"),
        GraphicFeature(start=1999500, end=2000000, strand=+1, color="#ccccff"),
    ],
)

ax, _ = record.plot(figure_width=6, x_lim=(1999000, None))
assert isinstance(ax.figure, plt.Figure)
ax.figure.savefig("sequence_and_translation.png", bbox_inches="tight")

Supposed change

just use ax.get_xlim() to get the actual plotted genome range.

        if is_undirected or head_is_cut:
            head_length = 0.001
        else:
            width_pixel = self._get_ax_width(ax, unit="pixel")
            head_length = 0.5 * width_pixel * feature.length / (ax.get_xlim()[1] - ax.get_xlim()[0])
            head_length = min(head_length, 0.6 * feature.thickness)

Thanks for your consideration!

[veghp]

Thank you for brining attention to this and for the proposed solution. I implemented the proposed change. The test outputs all looked fine.

Here is variant of the example code, for future reference:

n = 100000
s = 500

record = GraphicRecord(
    sequence=("N" * (n)),
    features=[
        GraphicFeature(start=n-2*s, end=n-s, strand=+1, color="#ffcccc"),
        GraphicFeature(start=n-s, end=n, strand=+1, color="#ccccff"),
    ],
)
ax, _ = record.plot(figure_width=6, x_lim=(n-2*s, None))

[Hocnonsense]

Thanks for your responce!

Here I also record another way to plot it, potentially more elegant in my case, sorry for my late reply:

n = 100000
s = 500

record = GraphicRecord(
    sequence=("N" * (n)),
    features=[
        GraphicFeature(start=n-2*s, end=n-s, strand=+1, color="#ffcccc"),
        GraphicFeature(start=n-s, end=n, strand=+1, color="#ccccff"),
    ],
)
ax, _ = record.crop(n-2*s, n).plot(figure_width=6)