def plot_sssh(
self,
feature_name: str,
feature_data: pandas.DataFrame,
top_feature_values: list,
stability_points: tuple = None,
ax: plt.axes = None,
color: str = 'YlGnBu',
stable_point: bool = False,
break_point: bool = False,
) -> None:
feature_data = feature_data.loc[
feature_data['feature_value'].isin(top_feature_values), :
]
feature_data = feature_data.assign(
feature_value=feature_data['feature_value'].astype(str)
)
feature_values = feature_data.feature_value.unique()
palette = sns.color_palette(
color, n_colors=self.top_feature_values
)[:len(feature_values)]
p2 = sns.lineplot(
data=feature_data, x='read', y='sssh', hue='feature_value',
ax=ax, ci=None, estimator=None, palette=palette,
hue_order=[str(v) for v in top_feature_values]
)
if stability_points is not None and stable_point:
ax.axvline(
x=stability_points[0], linewidth=1, linestyle='--', color='black'
)
if stability_points is not None and break_point:
if stability_points[1] is not None:
ax.axvline(
x=stability_points[1], linewidth=1, linestyle='-', color='black'
)
legend = ax.legend()
legend.texts[0].set_text(feature_name)
# Legend and labels
p2.tick_params(labelsize=6)
p2.set_xlabel('\nReads', fontsize=9)
p2.set_ylabel('Sum of ranked sum of shared hashes\n', fontsize=9)
def dist_attr(ax: plt.axes = None,
ds: loompy.LoomConnection = None,
out_file: str = None,
attr: str = None,
plot_title: str = None,
line: float = None):
if ax is None:
fig, ax = plt.subplots(figsize=(6, 4))
if attr in ds.ca:
ax.hist(ds.ca[attr], bins=100)
ax.set_xlim(np.amin(ds.ca[attr]), np.amax(ds.ca[attr]))
if line is not None:
ax.axvline(x=line, c='r')
if plot_title is not None:
ax.set_title(plot_title)
if out_file is not None:
fig.savefig(out_file, dpi=144)
return (ax)
def fake_doublets_dist(ax: plt.axes = None,
doublet_score_A: np.array = None,
logprob: np.array = None,
xx: np.array = None,
score1: float = 1,
score2: float = 1,
score: float = 1,
out_file: str = None) -> None:
# fig, ax = plt.subplots(figsize=(12, 12))
if ax is None:
ax = plt.gca()
ax.fill_between(xx.T[0], np.exp(logprob), alpha=0.5)
ax.plot(doublet_score_A,
np.full_like(doublet_score_A, -0.01),
'|k',
markeredgewidth=1)
ax.set_ylim(-0.02, 5)
ax.set_title(f'Fake Doublets distribution (Selected TH: {score})')
if score1 != 1:
ax.axvline(x=score1, c='r')
if score2 != 1:
ax.axvline(x=score2, linestyle='--', c='r')
if score != 1:
ax.axvline(x=score, linestyle=':', c='r')
ax.set_ylabel('# cells')
ax.set_xlabel('DoubletFinder score')
hd = ax.hist(doublet_score_A, bins=30, density=True)
if out_file is not None:
plt.savefig(out_file, dpi=144)
return (hd)
def plot_date_randomisation(ax: plt.axes,
replicates: np.array or list,
rate: float,
log10: bool = True) -> plt.axes:
""" Plot distribution of substitution rates for date randomisation test
:param ax: axes object to plot the date randomisation
:param replicates: list of replicate substitution rate estimates
:param rate: true rate estimate vertical line for evaluation
:param log10: plot log10 of substitution rates on horizontal axis
:returns axes object
"""
if log10:
replicates = np.log10(replicates)
with plt.style.context('seaborn-colorblind'):
ax.hist(x=replicates, color='gray')
ax.axvline(x=rate, color='r')
return ax
