def _plot_data(self):
df = densiometry.calc_mean_change(self.df)
df = densiometry.sort_by_activity(df, self.select_designs)
if self.designs:
designs = self.designs
elif self.show_all_designs:
designs = self.all_designs
else:
designs = self.select_designs
data = np.vstack(
[df[df.design == x].mean_change.values for x in designs])
spacers = df['spacer'].unique()
num_spacers = len(spacers)
spacers_per_row = int(np.ceil(num_spacers / self.rows))
for i in range(self.rows):
j0 = spacers_per_row * (i + 0)
j1 = spacers_per_row * (i + 1)
self.ax[i].pcolor(data[:, j0:j1], cmap=self.cmap, norm=self.norm)
self.ax[i].set_yticks(0.5 + np.arange(len(self.designs)))
self.ax[i].set_yticklabels(self.designs)
self.ax[i].set_xticks(0.5 + np.arange(spacers_per_row))
self.ax[i].set_xticklabels(
spacers[j0:j1] if self.label_spacers else [])
self.ax[i].axis('image')
def _plot_data(self):
# For dot plots:
dots = densiometry.calc_percent_change(self.df)
# For bar plots:
bars = densiometry.calc_mean_change(self.df)
bars = densiometry.sort_by_activity(bars, self.designs)
for i, design in enumerate(self.designs):
x_from_spacer = {}
success_count = 0
for x, row in bars[bars.design == design].reset_index().iterrows():
x_from_spacer[row.spacer] = x
success_count += self._plot_bar(
self.axes[i],
x,
row.mean_change,
None, # Don't plot error bars, since we're plotting every point.
self.colors[design],
)
dots_i = dots.xs(design, level='design')
for spacer, group in dots_i.groupby(['spacer']):
self._plot_dots(
self.axes[i],
x_from_spacer[spacer],
group.percent_change.values,
self.dot_colors[design],
)
# If a figure size was given, also try to take some steps to ensure
# a more regular/representative size. In particular, use shorter
# labels and don't label redundant axes.
y_max = self.y_max[design]
self.axes[i].set_ylabel(
'Δ cleavage (%)' if not self.fig_size else 'Δ (%)')
self.axes[i].set_ylim((0, y_max))
self.axes[i].set_yticks([0, y_max // 2, y_max])
self.axes[i].set_xlabel('spacer' if i == 1 else '')
self.axes[i].set_xlim((-3, x + 1))
self.axes[i].set_xticks(range(-2, x + 1))
self.axes[i].set_xticklabels(['on', 'off'] +
[str(x) for x in range(1, x + 2)],
rotation='vertical')
self.axes[i].axhline(
self.success_cutoff,
color=ucsf.light_grey[0],
linestyle='--',
)
print(f'{design}: {success_count}/{x+1}')
def load_rational_designs():
df = densiometry.load_cleavage_data_from_xlsx_dir('densiometry')
df = densiometry.calc_mean_change(df)
df = df.apply(name_to_strategy, axis='columns')
df['strategy'] = pd.Categorical(df.strategy, [e.value for e in Strategy])
df['domain'] = pd.Categorical(df.domain, [e.value for e in Domain])
df['active'] = abs(df.mean_change) > 0.15
df['sequence'] = df.apply(name_to_sequence, axis='columns')
#df['aligned_sequence'] = align_sequences(df.sequence)
df = discard_modified_aptamers(df)
df = get_pretty_sequences(df)
priority = [
'align_group', 'strategy', 'algorithm', 'domain', 'earliest_date',
'earliest_order'
]
df = df.sort_values(priority).reset_index(drop=True)
df['id'] = df.index.values + 1
return df
def get_construct(x):
names = {
'on': 'pos',
'off': 'neg',
'mhf 30': r'\ligrnaF{}',
'rxb 11,1': r'\ligrnaB{}',
}
return names[x]
args = docopt.docopt(__doc__)
# Parse the densiometry data.
df = densiometry.load_cleavage_data_from_xlsx_dir(DATA_DIR)
df = densiometry.calc_mean_change(df)
# Add NaNs for the experiments that didn't work.
df = df.append({'spacer': 'd9', 'design': 'on'}, ignore_index=True)
# Add a column for the names I want to display.
df = df[df.design.isin(DESIGN_ORDER)]
df['construct'] = df.design.apply(get_construct)
# Put the rows in the order I want to display them in.
spacer_order = densiometry.sort_spacers_by_activity(df, ['mhf 30', 'rxb 11,1'])
df['spacer'] = pd.Categorical(df.spacer, spacer_order)
df['design'] = pd.Categorical(df.design, DESIGN_ORDER)
df = df.sort_values(['spacer', 'design']).reset_index()
n = len(df) // 2