def similarity_histogram(groups,
mode='j',
bins=10,
title=None,
filename=None,
figsize=None):
'''
Given a group of sets, generates a histogram of the similarity
indices across each possible pair (same-element pairs excluded).
* Arguments:
- *groups* [list]: Or any iterable of [set] objects.
- *mode* [str]: Optional, ``'j'`` (Jaccard) by default.
Indicates which type of similarity index/coefficient is to be
computed. Available options are: ``'j'`` for Jaccard, ``'sd'``
for Sorensen-Dice and ``'ss'`` for Szymkiewicz–Simpson. See
:py:func:`data_tools.iterables.similarity` for more
information.
- *bins* [int]: Optional, ``10`` by default. Number of bins to
show in the histogram.
- *title* [str]: Optional, ``None`` by default. Defines the plot
title.
- *filename* [str]: Optional, ``None`` by default. If passed,
indicates the file name or path where to store the figure.
Format must be specified (e.g.: .png, .pdf, etc)
- *figsize* [tuple]: Optional, ``None`` by default (default
matplotlib size). Any iterable containing two values denoting
the figure size (in inches) as [width, height].
* Returns:
- [matplotlib.figure.Figure]: The figure object containing a
combination of box and scatter plots of the gene-set scores,
unless *filename* is provided.
'''
sims = [
similarity(a, b, mode=mode)
for (a, b) in itertools.combinations_with_replacement(groups, 2)
]
fig, ax = plt.subplots(figsize=figsize)
ax.hist(sims, bins=bins)
ax.set_xlabel('Similarity index')
ax.set_ylabel('Frequency')
ax.set_title(title)
fig.tight_layout()
if filename:
fig.savefig(filename)
else:
return fig
def similarity_heatmap(groups, labels=None, mode='j', cmap='nipy_spectral',
title=None, filename=None, figsize=None):
'''
Given a group of sets, generates a heatmap with the similarity
indices across each possible pair.
* Arguments:
- *groups* [list]: Or any iterable of [set] objects.
- *labels* [list]: Optional, ``None`` by default. Labels for the
sets following the same order as provided in *groups*.
- *mode* [str]: Optional, ``'j'`` (Jaccard) by default.
Indicates which type of similarity index/coefficient is to be
computed. Available options are: ``'j'`` for Jaccard, ``'sd'``
for Sorensen-Dice and ``'ss'`` for Szymkiewicz–Simpson. See
:py:func:`data_tools.iterables.similarity` for more
information.
- *cmap* [str]: Optional, ``'nipy_spectral'`` by default. The
colormap used for the plot (can also be a
[matplotlib.colors.Colormap] object). See other [str] options
available in `Matplotlib's reference manual`_.
- *title* [str]: Optional, ``None`` by default. Defines the plot
title.
- *filename* [str]: Optional, ``None`` by default. If passed,
indicates the file name or path where to store the figure.
Format must be specified (e.g.: .png, .pdf, etc)
- *figsize* [tuple]: Optional, ``None`` by default (default
matplotlib size). Any iterable containing two values denoting
the figure size (in inches) as [width, height].
.. _`Matplotlib's reference manual`:
https://matplotlib.org/examples/color/colormaps_reference.html
* Returns:
- [matplotlib.figure.Figure]: The figure object containing a
combination of box and scatter plots of the gene-set scores,
unless *filename* is provided.
'''
sims = []
for (a, b) in itertools.product(groups, repeat=2):
sims.append(similarity(set(a), set(b), mode=mode))
# Convert similarity indices to square matrix
sims = np.array(sims).reshape(len(groups), len(groups))
# Plotting heatmap for a given similarity index
fig, ax = plt.subplots(figsize=figsize)
im = ax.imshow(sims, cmap=cmap, interpolation='none')
fig.colorbar(im)
if labels:
try:
a, b = map(len, [groups, labels])
assert a == b
except AssertionError as e:
raise e('Invalid length of labels %d != %d' % (a, b))
rng = range(len(groups))
ax.set_xticks(rng)
ax.set_xticklabels(labels, rotation=90)
ax.set_yticks(rng)
ax.set_yticklabels(labels)
ax.set_title(title)
fig.tight_layout()
if filename:
fig.savefig(filename)
else:
return fig