def create_mask(data, use_sig=True, p_value=0.1, fold_change_cutoff=1.5):
""" Creates a mask for volcano plots.
# Visual example of volcano plot
# section 0 are significant criteria
# 0 # 1 # 0 #
# # # #
#################################
# # # #
# 2 # 2 # 2 #
# # # #
#################################
Parameters
----------
data : pd.DataFrame
use_sig : bool
p_value : float
p_value threshold
fold_change_cutoff : float
fold change threshold
Returns
-------
"""
# copy of data
tmp = data.loc[:, (p_val, fold_change, flag)].copy()
# convert to log10 scale
tmp[p_val] = np.log10(data[p_val]) * -1
# convert to log2 space
tmp = log2_normalize_df(tmp, column=fold_change)
if use_sig:
sec_0 = tmp[tmp[flag]]
sec_2 = tmp[~tmp[flag]]
sec_1 = None
else:
fc = np.log2(fold_change_cutoff)
p_value = -1 * np.log10(p_value)
criteria_1 = tmp[p_val] >= p_value
sec_0 = tmp[criteria_1 & (np.abs(tmp[fold_change]) >= fc)]
sec_1 = tmp[criteria_1 & (np.abs(tmp[fold_change]) < fc)]
sec_2 = tmp[(tmp[p_val] < p_value)]
return sec_0, sec_1, sec_2
def create_histogram_measurements(self,
exp_data_type,
save_name=None,
y_range=None,
out_dir=None):
"""
Plots a histogram of data
Parameters
----------
exp_data_type: str
Which data to plot
save_name: str
Name of figure
out_dir: str, path
Path to location to save figure
y_range: array_like
range of data
Returns
-------
"""
if not self._check_experiment_type_existence(exp_type=exp_data_type):
return
data = self.data[self.data[exp_method] == exp_data_type].copy()
data = data.dropna(subset=[p_val])
data = data[np.isfinite(data[fold_change])]
data = data.dropna(subset=[fold_change])
tmp = np.array(log2_normalize_df(data, fold_change)[fold_change])
fig = plt.figure()
ax = fig.add_subplot(111)
ax.hist(tmp, 50, color='gray')
if y_range is not None:
plt.xlim(y_range[0], y_range[1])
ax.set_yscale('log', basey=10)
ax.set_xlabel('log$_2$ Fold Change', fontsize=16)
ax.set_ylabel('Count', fontsize=16)
if save_name is not None:
v_plot.save_plot(fig, save_name, out_dir)
return fig
def create_histogram_measurements(self, save_name=None,
y_range=None, out_dir=None):
"""
Plots a histogram of data
Parameters
----------
save_name: str
Name of figure
out_dir: str, path
Path to location to save figure
y_range: array_like
range of data
Returns
-------
"""
data = self.copy()
data = data.dropna(subset=[p_val])
data = data[np.isfinite(data[fold_change])]
data = data.dropna(subset=[fold_change])
tmp = np.array(log2_normalize_df(data, fold_change)[fold_change])
fig = plt.figure()
ax = fig.add_subplot(111)
ax.hist(tmp, 50, color='gray')
if y_range is not None:
plt.xlim(y_range[0], y_range[1])
ax.set_yscale('log', basey=10)
ax.set_xlabel('log$_2$ Fold Change', fontsize=16)
ax.set_ylabel('Count', fontsize=16)
fig.tight_layout()
if save_name is not None:
v_plot.save_plot(fig, save_name, out_dir)
return fig
def plot_histogram(self, save_name=None, y_range=None, out_dir=None):
"""
Plots a histogram of data
Parameters
----------
save_name: str
Name of figure
out_dir: str, path
Path to location to save figure
y_range: array_like
range of data
Returns
-------
"""
data = self.copy()
data = data.dropna(subset=[p_val])
data = data[np.isfinite(data[fold_change])]
data = data.dropna(subset=[fold_change])
tmp = np.array(log2_normalize_df(data, fold_change)[fold_change])
fig = plt.figure()
ax = fig.add_subplot(111)
ax.hist(tmp, 50, color='gray')
if y_range is not None:
plt.xlim(y_range[0], y_range[1])
ax.set_yscale('log', basey=10)
ax.set_xlabel('log$_2$ Fold Change', fontsize=16)
ax.set_ylabel('Count', fontsize=16)
fig.tight_layout()
if save_name is not None:
v_plot.save_plot(fig, save_name, out_dir)
return fig
def plot_species(df, species_list=None, save_name='test', out_dir=None,
title=None, plot_type='plotly', image_format='pdf',
close_plots=False):
"""
Parameters
----------
df: pandas.DataFrame
magine formatted dataframe
species_list: list
List of genes to be plotter
save_name: str
Filename to be saved as
out_dir: str
Path for output to be saved
title: str
Title of plot, useful when list of genes corresponds to a GO term
plot_type : str
Use plotly to generate html output or matplotlib to generate pdf
image_format : str
pdf or png, only used if plot_type="matplotlib"
close_plots : bool
Close plot after making, use when creating lots of plots in parallel.
Returns
-------
"""
ldf = df.copy()
if out_dir is not None:
if not os.path.exists(out_dir):
os.mkdir(out_dir)
# gather x axis points
x_points = sorted(ldf[sample_id].unique())
if len(x_points) == 0:
return
if isinstance(x_points[0], np.float):
x_point_dict = {i: x_points[n] for n, i
in enumerate(x_points)}
else:
x_point_dict = {i: n for n, i
in enumerate(x_points)}
if species_list is not None:
ldf = ldf.loc[ldf[identifier].isin(species_list)].copy()
ldf = log2_normalize_df(ldf, column=fold_change)
n_plots = len(ldf[identifier].unique())
num_colors = len(ldf[label_col].unique())
color_list = sns.color_palette("tab20", num_colors)
if plot_type == 'matplotlib':
fig = plt.figure()
ax = fig.add_subplot(111)
ax.set_prop_cycle(plt.cycler('color', color_list))
colors = enumerate(color_list)
plotly = []
names_list = []
total_counter = 0
for name, j in ldf.groupby(identifier):
index_counter = 0
for n, m in j.groupby(label_col):
x = np.array(m[sample_id])
if len(x) < 1:
continue
y = np.array(m['fold_change'])
sig_flag = np.array(m[flag])
index = np.argsort(x)
x = x[index]
y = y[index]
s_flag = sig_flag[index]
# x values with scaled values (only changes things if non-float
# values are used for sample_id
x_index = np.array([x_point_dict[ind] for ind in x])
index_counter += 1
total_counter += 1
# create matplotlib plot
if plot_type == 'matplotlib':
label = "\n".join(wrap(n, 40))
p = ax.plot(x_index, y, '.-', label=label)
if len(s_flag) != 0:
color = p[0].get_color()
ax.plot(x_index[s_flag], y[s_flag], '^', color=color)
# create plotly plot
elif plot_type == 'plotly':
c = next(colors)[1]
plotly.append(_ploty_graph(x_index, y, n, n, c))
if len(s_flag) != 0:
index_counter += 1
total_counter += 1
plotly.append(_ploty_graph(x_index[s_flag], y[s_flag],
n, n, c, marker='x-open-dot'))
names_list.append([name, index_counter])
if plot_type == 'matplotlib':
lgd = _format_mpl(ax, x_point_dict, x_points)
if save_name is not None:
tmp_savename = "{}.{}".format(save_name, image_format)
if out_dir is not None:
tmp_savename = os.path.join(out_dir, tmp_savename)
plt.savefig(tmp_savename, bbox_extra_artists=(lgd,),
bbox_inches='tight')
if close_plots:
plt.close(fig)
else:
return fig
elif plot_type == 'plotly':
fig = _create_plotly(total_counter, n_plots, names_list, x_point_dict,
title, x_points, plotly)
if save_name:
_save_ploty_output(fig, out_dir, save_name)
else:
init_notebook_mode(connected=True)
iplot(fig)
def plot_species(df, species_list=None, save_name='test', out_dir=None,
title=None, plot_type='plotly', image_format='pdf',
close_plots=False):
"""
Parameters
----------
df: pandas.DataFrame
magine formatted dataframe
species_list: list
List of genes to be plotter
save_name: str
Filename to be saved as
out_dir: str
Path for output to be saved
title: str
Title of plot, useful when list of genes corresponds to a GO term
plot_type : str
Use plotly to generate html output or matplotlib to generate pdf
image_format : str
pdf or png, only used if plot_type="matplotlib"
close_plots : bool
Close plot after making, use when creating lots of plots in parallel.
Returns
-------
"""
ldf = df.copy()
if out_dir is not None:
if not os.path.exists(out_dir):
os.mkdir(out_dir)
# gather x axis points
x_points = sorted(ldf[sample_id].unique())
if len(x_points) == 0:
return
if isinstance(x_points[0], np.float):
x_point_dict = {i: x_points[n] for n, i
in enumerate(x_points)}
else:
x_point_dict = {i: n for n, i
in enumerate(x_points)}
if species_list is not None:
ldf = ldf.loc[ldf[identifier].isin(species_list)].copy()
ldf = log2_normalize_df(ldf, column=fold_change)
n_plots = len(ldf[identifier].unique())
num_colors = len(ldf[label_col].unique())
color_list = sns.color_palette("tab20", num_colors)
if plot_type == 'matplotlib':
fig = plt.figure()
ax = fig.add_subplot(111)
ax.set_prop_cycle(plt.cycler('color', color_list))
colors = enumerate(color_list)
plotly = []
names_list = []
total_counter = 0
for name, j in ldf.groupby(identifier):
index_counter = 0
for n, m in j.groupby(label_col):
x = np.array(m[sample_id])
if len(x) < 1:
continue
y = np.array(m['fold_change'])
sig_flag = np.array(m[flag])
index = np.argsort(x)
x = x[index]
y = y[index]
s_flag = sig_flag[index]
# x values with scaled values (only changes things if non-float
# values are used for sample_id
x_index = np.array([x_point_dict[ind] for ind in x])
index_counter += 1
total_counter += 1
# create matplotlib plot
if plot_type == 'matplotlib':
label = "\n".join(wrap(n, 40))
p = ax.plot(x_index, y, '.-', label=label)
if len(s_flag) != 0:
color = p[0].get_color()
ax.plot(x_index[s_flag], y[s_flag], '^', color=color)
# create plotly plot
elif plot_type == 'plotly':
c = next(colors)[1]
plotly.append(_ploty_graph(x_index, y, n, n, c))
if len(s_flag) != 0:
index_counter += 1
total_counter += 1
plotly.append(_ploty_graph(x_index[s_flag], y[s_flag],
n, n, c, marker='x-open-dot'))
names_list.append([name, index_counter])
if plot_type == 'matplotlib':
lgd = _format_mpl(ax, x_point_dict, x_points)
if save_name is not None:
tmp_savename = "{}.{}".format(save_name, image_format)
if out_dir is not None:
tmp_savename = os.path.join(out_dir, tmp_savename)
plt.savefig(tmp_savename, bbox_extra_artists=(lgd,),
bbox_inches='tight')
if close_plots:
plt.close(fig)
else:
return fig
elif plot_type == 'plotly':
fig = _create_plotly(total_counter, n_plots, names_list, x_point_dict,
title, x_points, plotly)
if save_name:
_save_ploty_output(fig, out_dir, save_name)
else:
init_notebook_mode(connected=True)
iplot(fig)