def plotProjection(results, axis1=0, axis2=1,
names_list=CN_globalVars.sm_names, scolor_dict=CN_globalVars.scolors, markers_dict=CN_globalVars.markers,
leaveSM=None, leaveSM_results=None, leaveSM_edgeColor='orange',
figsize=(6.8, 6.5), axisLabelPrefix='', title=None, bbox_to_anchor=(1.13, 0.875)):
fig = plt.figure(figsize=figsize)
ax = plt.subplot(111)
i = 0
for currSM in names_list:
if currSM != leaveSM:
ax.scatter(results[i, axis1], results[i, axis2],
color=scolor_dict[currSM], marker=markers_dict[currSM], label=currSM,
edgecolor='black', linewidth=2, s=120)
i += 1
if leaveSM:
ax.scatter(leaveSM_results[0, axis1], leaveSM_results[0, axis2],
color=scolor_dict[leaveSM], marker=markers_dict[leaveSM], label=leaveSM,
edgecolor=leaveSM_edgeColor, linewidth=2, s=120)
plotlib.setproperties(xlabel=axisLabelPrefix+'component {}'.format(axis1+1),
ylabel=axisLabelPrefix+'component {}'.format(axis2+1),
title=title,
labelfontsize=22, tickfontsize=22,
symmetric=True, tight=False)
ax.legend(bbox_to_anchor=bbox_to_anchor, bbox_transform=fig.transFigure, fontsize=14)
return ax
def plotBindingCurvesAcrossTargets(affinityDataList,
variant,
colors=None,
names=None,
norm=False,
ax=None):
'''Plot a binding curve of a particular variant'''
if colors is None:
colors = sns.color_palette("Paired", 12)
if ax is None:
fig = plt.figure(figsize=(4, 4))
ax = fig.add_subplot(111)
for i, affinityData in enumerate(affinityDataList):
# Load data
subSeries = affinityData.getVariantBindingSeries(variant)
if len(subSeries) == 0:
print 'No fluorescence data associated with variant %s' % str(
variant)
return
concentrations = affinityData.x
variant_table = affinityData.variant_table
variant_data = variant_table.loc[variant]
if norm:
fmax = variant_data['fmax']
subSeries = subSeries.copy() / fmax
variant_data = variant_data.copy()
for field in [
'fmax_init', 'fmin_init', 'fmax_lb', 'fmax', 'fmax_ub',
'fmin_lb', 'fmin', 'fmin_ub'
]:
variant_data[field] = variant_data[field] / fmax
# Plot
plotting.plotFitCurve(concentrations, subSeries, variant_data, ax=ax)
ax.lines[i * 4 + 3].set_color(colors[i])
ax.lines[i * 4 + 3].set_linewidth(2)
if names is not None:
ax.lines[i * 4 + 3].set_label(names[i])
if norm:
ylabel = 'Normalized fluorescence'
else:
ylabel = 'Fluorescence'
plotlib.setproperties(xlabel='Concentration (nM)',
ylabel=ylabel,
labelfontsize=16,
tickfontsize=16,
legend=names,
legendloc=2,
legendfontsize=12)
return ax
def doubleMutant(allData_aggPF, mutantSM, targetSM, normSM, names, norm=False, coop=False):
# Define variables
data = allData_aggPF[targetSM]['sm_norm_diff_signals_2']['median']
mutantRefVariant = allData_aggPF[mutantSM]['sm_norm_diff_signals_2']['median'].idxmax()
targetRefVariant = allData_aggPF[targetSM]['sm_norm_diff_signals_2']['median'].idxmax()
normRefVariant = allData_aggPF[normSM]['sm_norm_diff_signals_2']['median'].idxmax()
if norm:
targetRefSignal = allData_aggPF[targetSM]['sm_norm_diff_signals_2']['median'][targetRefVariant]
normRefSignal = allData_aggPF[normSM]['sm_norm_diff_signals_2']['median'][normRefVariant]
else:
targetRefSignal = 1
normRefSignal = 1
libSeq = 'NNGGATTTTCCNNNNACGAAGTNNTCCCGAG'
startPos = 14
title = 'Mutants of '+names[mutantSM].lower()+'0, in the presence of '+names[targetSM]
if norm and coop:
cbarLabel = 'Cooperativity normalized to '+names[targetSM].lower()+'0'
elif norm and not coop:
cbarLabel = 'Differential signals normalized to '+names[targetSM].lower()+'0'
elif not norm and coop:
cbarLabel = 'Cooperativity'
else:
cbarLabel = 'Differential signals'
# Compute vmin and vmax manually (assuming robust=True) so that the color bar
# scale is the same across the same dataRefVariant
normRefData = allData_aggPF[normSM]['sm_norm_diff_signals_2']['median'] / normRefSignal
doubleMutantSignals, mutantLabels = plotlib.doubleMutantMatrix(normRefData, normRefVariant, libSeq, startPos, coop)
doubleMutantSignals = doubleMutantSignals[~np.isnan(doubleMutantSignals)]
vmin = np.percentile(doubleMutantSignals, 1)
vmax = np.percentile(doubleMutantSignals, 99)
vlim = max(abs(vmin), abs(vmax))
vmin, vmax = -vlim, vlim
# Make heatmap
plt.figure(figsize=(12,10))
ax, cax = plotlib.doubleMutant(data, mutantRefVariant, libSeq,
startPos=startPos, refSignal=targetRefSignal,
normToRefSignal=norm, coop=coop,
vmin=vmin, vmax=vmax, cbarLabel=cbarLabel,
triangle='lower', invertY=False, linewidth=3)
cax.tick_params(labelsize=20)
cax.yaxis.label.set_fontsize(20)
plotlib.setproperties(title=title, labelfontsize=20, titlefontsize=20)
# Save to file
if coop:
figureOutput = 'coop'
else:
figureOutput = 'doubleMutant'
if norm:
figureNorm = 'norm_'
else:
figureNorm = ''
plt.savefig(figureDir+'/'+names[targetSM]+'_'+names[mutantSM].lower()+'0_'+names[normSM].lower()+'0_'+figureNorm+figureOutput+'.png')
return ax, cax
def plotScatterNorm_ds2_bserr(data_aggPF, name, fullName):
x = data_aggPF['sm_norm_diff_signals_2']['median']
y = data_aggPF['sm_norm_diff_signals_2']['bserr'].apply(lambda x: float(x.split(',')[1][:-1]) - float(x.split(',')[0][1:]))
plt.figure(figsize=(7, 7))
plotlib.scatterDensity(x, y, alpha=0.7, log=True)
plotlib.setproperties(xlabel='Normalized differential signals',
ylabel='Bootstrapped error',
title=fullName,
labelfontsize=25, tickfontsize=25, equal=True)
plt.savefig(figureDir+'/'+name+'_scatterNorm_ds2_bserr.png')
def plotBindingCurve(affinityData, variant, annotate=True, ax=None):
'''Plot a binding curve of a particular variant'''
# Load data
subSeries = affinityData.getVariantBindingSeries(variant)
if len(subSeries) == 0:
print 'No fluorescence data associated with variant %s' % str(variant)
return
concentrations = affinityData.x
variant_table = affinityData.variant_table
# Plot
if ax is None:
fig = plt.figure(figsize=(4, 4))
ax = fig.add_subplot(111)
plotting.plotFitCurve(concentrations,
subSeries,
variant_table.loc[variant],
ax=ax)
plotlib.setproperties(xlabel='Concentration (nM)',
ylabel='Fluorescence',
labelfontsize=16,
tickfontsize=16)
# Annonate
if annotate:
names = [
'dG', 'dG_lb', 'dG_ub', 'fmax', 'fmax_lb', 'fmax_ub', 'numTests',
'pvalue', 'rsq', 'flag'
]
vec = pd.Series([
processresults.getValueInTable(variant_table.loc[variant], name)
for name in names
],
index=names)
#vec.fillna('', inplace=True)
annotationText = [
'Kd = {:4.2f} uM ({:4.2f}, {:4.2f})'.format(
liblib.dGtoKd(vec.dG, 'uM'), liblib.dGtoKd(vec.dG_lb, 'uM'),
liblib.dGtoKd(vec.dG_ub, 'uM')),
'fmax = {:4.2f} ({:4.2f}, {:4.2f})'.format(vec.fmax, vec.fmax_lb,
vec.fmax_ub),
'Nclusters = {:4.0f}'.format(vec.numTests),
'pvalue = {:.1e}'.format(vec.pvalue),
'average Rsq = {:4.2f}'.format(vec.rsq),
'fmax enforced = {}'.format(vec.flag)
]
ax.annotate('\n'.join(annotationText),
xy=(0.05, .95),
xycoords='axes fraction',
horizontalalignment='left',
verticalalignment='top',
fontsize=11)
return ax
def plotScatterNorm_ds2_fbserr(data_aggPF, name, fullName):
x = data_aggPF['sm_norm_diff_signals_2']['median']
y = data_aggPF['sm_norm_diff_signals_2']['bserr'].apply(lambda x: float(x.split(',')[1][:-1]) - float(x.split(',')[0][1:]))
df = pd.concat([x, y / x], axis=1).replace([np.inf, -np.inf], np.nan).dropna()
plt.figure(figsize=(7, 7))
plotlib.scatterDensity(df[0], df[1], alpha=0.7, log=True)
plotlib.setproperties(xlabel='Normalized differential signals',
ylabel='Fractional bootstrapped error',
title=fullName,
labelfontsize=25, tickfontsize=25, equal=True)
plt.savefig(figureDir+'/'+name+'_scatterNorm_ds2_fbserr.png')
def plotScatterNorm_QO_ds2(data_aggPF, name, fullName):
x = data_aggPF['QO_norm_signals']['median']
y = data_aggPF['sm_norm_diff_signals_2']['median']
plt.figure(figsize=(7, 7))
plotlib.scatterDensity(x, y, alpha=0.7, log=True)
plotlib.setproperties(xlim=(-0.09, 1), ylim=(-0.09, 1),
xlabel='Normalized quenched signals',
ylabel='Normalized differential signals 2',
title=fullName,
labelfontsize=25, tickfontsize=25, equal=True)
plt.savefig(figureDir+'/'+name+'_scatterNorm_QO_ds2.png')
def plotScatterNorm_QO_sm_linear(data_aggPF, name, fullName):
x = data_aggPF['quenched_norm']['median']
y = data_aggPF['switch2_norm']['median']
plt.figure(figsize=(7, 7))
plotlib.scatterDensity(x, y, alpha=0.7, log=False)
plotlib.setproperties(xlim=(-0.09, 1), ylim=(-0.09, 1),
xlabel='Normalized quenched signals',
ylabel='Normalized switching signals',
title=fullName,
labelfontsize=25, tickfontsize=25, equal=True)
plt.savefig(figureDir+'/'+name+'_scatterNorm_QO_sm_linear.png')
def plotConfusionMatrix(trueY, predictedY, catLabels=None, figsize=(8, 7)):
fig = plt.figure(figsize=figsize)
ax = sns.heatmap(confusion_matrix(trueY, predictedY),
annot=True, square=True,
xticklabels=catLabels, yticklabels=catLabels)
cax = fig.axes[-1]
plotlib.setproperties(ax=ax, fontsize=15, yticklabelrot=0,
xlabel='Predicted label', ylabel='True label')
plotlib.setproperties(ax=cax, fontsize=15, yticklabelrot=0)
return ax
def plotScatter_QO_sm(data_aggPF, name, fullName):
x = data_aggPF['QO_signals']['median']
y = data_aggPF['sm_signals']['median']
plt.figure(figsize=(7, 7))
plotlib.scatterDensity(x, y, alpha=0.7, log=True)
ub = max(np.percentile(x, 99.9), np.percentile(y, 99.9)) * 1.1
plotlib.setproperties(xlim=(0, ub), ylim=(0, ub),
xlabel='Quenched signals',
ylabel='Switching signals',
title=fullName,
labelfontsize=25, tickfontsize=25, equal=True)
plt.savefig(figureDir+'/'+name+'_scatter_QO_sm.png')
def plotScatterNorm_ds2_bserr(data_aggPF, name, fullName):
x = data_aggPF['sm_norm_diff_signals_2']['median']
y = data_aggPF['sm_norm_diff_signals_2']['bserr'].apply(
lambda x: float(x.split(',')[1][:-1]) - float(x.split(',')[0][1:]))
plt.figure(figsize=(7, 7))
plotlib.scatterDensity(x, y, alpha=0.7, log=True)
plotlib.setproperties(xlabel='Normalized differential signals',
ylabel='Bootstrapped error',
title=fullName,
labelfontsize=25,
tickfontsize=25,
equal=True)
plt.savefig(figureDir + '/' + name + '_scatterNorm_ds2_bserr.png')
def plotScatterNorm_QO_ds1(data_aggPF, name, fullName):
x = data_aggPF['QO_norm_signals']['median']
y = data_aggPF['sm_norm_diff_signals_1']['median']
plt.figure(figsize=(7, 7))
plotlib.scatterDensity(x, y, alpha=0.7, log=True)
plotlib.setproperties(xlim=(-0.09, 1),
ylim=(-0.09, 1),
xlabel='Normalized quenched signals',
ylabel='Normalized differential signals 1',
title=fullName,
labelfontsize=25,
tickfontsize=25,
equal=True)
plt.savefig(figureDir + '/' + name + '_scatterNorm_QO_ds1.png')
def plotProjection(results,
axis1=0,
axis2=1,
names_list=CN_globalVars.sm_names,
scolor_dict=CN_globalVars.scolors,
markers_dict=CN_globalVars.markers,
leaveSM=None,
leaveSM_results=None,
leaveSM_edgeColor='orange',
figsize=(6.8, 6.5),
axisLabelPrefix='',
title=None,
bbox_to_anchor=(1.13, 0.875)):
fig = plt.figure(figsize=figsize)
ax = plt.subplot(111)
i = 0
for currSM in names_list:
if currSM != leaveSM:
ax.scatter(results[i, axis1],
results[i, axis2],
color=scolor_dict[currSM],
marker=markers_dict[currSM],
label=currSM,
edgecolor='black',
linewidth=2,
s=120)
i += 1
if leaveSM:
ax.scatter(leaveSM_results[0, axis1],
leaveSM_results[0, axis2],
color=scolor_dict[leaveSM],
marker=markers_dict[leaveSM],
label=leaveSM,
edgecolor=leaveSM_edgeColor,
linewidth=2,
s=120)
plotlib.setproperties(
xlabel=axisLabelPrefix + 'component {}'.format(axis1 + 1),
ylabel=axisLabelPrefix + 'component {}'.format(axis2 + 1),
title=title,
labelfontsize=22,
tickfontsize=22,
symmetric=True,
tight=False)
ax.legend(bbox_to_anchor=bbox_to_anchor,
bbox_transform=fig.transFigure,
fontsize=14)
return ax
def plotScatter_QO_sm(data_aggPF, name, fullName):
x = data_aggPF['QO_signals']['median']
y = data_aggPF['sm_signals']['median']
plt.figure(figsize=(7, 7))
plotlib.scatterDensity(x, y, alpha=0.7, log=True)
ub = max(np.percentile(x, 99.9), np.percentile(y, 99.9)) * 1.1
plotlib.setproperties(xlim=(0, ub),
ylim=(0, ub),
xlabel='Quenched signals',
ylabel='Switching signals',
title=fullName,
labelfontsize=25,
tickfontsize=25,
equal=True)
plt.savefig(figureDir + '/' + name + '_scatter_QO_sm.png')
def plotScatterNorm_ds2_fbserr(data_aggPF, name, fullName):
x = data_aggPF['sm_norm_diff_signals_2']['median']
y = data_aggPF['sm_norm_diff_signals_2']['bserr'].apply(
lambda x: float(x.split(',')[1][:-1]) - float(x.split(',')[0][1:]))
df = pd.concat([x, y / x], axis=1).replace([np.inf, -np.inf],
np.nan).dropna()
plt.figure(figsize=(7, 7))
plotlib.scatterDensity(df[0], df[1], alpha=0.7, log=True)
plotlib.setproperties(xlabel='Normalized differential signals',
ylabel='Fractional bootstrapped error',
title=fullName,
labelfontsize=25,
tickfontsize=25,
equal=True)
plt.savefig(figureDir + '/' + name + '_scatterNorm_ds2_fbserr.png')
def plotBindingCurvesAcrossTargets(affinityDataList, variant,
colors=None, names=None, norm=False, ax=None):
'''Plot a binding curve of a particular variant'''
if colors is None:
colors = sns.color_palette("Paired", 12)
if ax is None:
fig = plt.figure(figsize=(4, 4))
ax = fig.add_subplot(111)
for i, affinityData in enumerate(affinityDataList):
# Load data
subSeries = affinityData.getVariantBindingSeries(variant)
if len(subSeries)==0:
print 'No fluorescence data associated with variant %s'%str(variant)
return
concentrations = affinityData.x
variant_table = affinityData.variant_table
variant_data = variant_table.loc[variant]
if norm:
fmax = variant_data['fmax']
subSeries = subSeries.copy() / fmax
variant_data = variant_data.copy()
for field in ['fmax_init', 'fmin_init',
'fmax_lb', 'fmax', 'fmax_ub',
'fmin_lb', 'fmin', 'fmin_ub']:
variant_data[field] = variant_data[field] / fmax
# Plot
plotting.plotFitCurve(concentrations, subSeries, variant_data, ax=ax)
ax.lines[i*4+3].set_color(colors[i])
ax.lines[i*4+3].set_linewidth(2)
if names is not None:
ax.lines[i*4+3].set_label(names[i])
if norm:
ylabel = 'Normalized fluorescence'
else:
ylabel = 'Fluorescence'
plotlib.setproperties(xlabel='Concentration (nM)', ylabel=ylabel,
labelfontsize=16, tickfontsize=16,
legend=names, legendloc=2, legendfontsize=12)
return ax
def plotConfusionMatrix(trueY, predictedY, catLabels=None, figsize=(8, 7)):
fig = plt.figure(figsize=figsize)
ax = sns.heatmap(confusion_matrix(trueY, predictedY),
annot=True,
square=True,
xticklabels=catLabels,
yticklabels=catLabels)
cax = fig.axes[-1]
plotlib.setproperties(ax=ax,
fontsize=15,
yticklabelrot=0,
xlabel='Predicted label',
ylabel='True label')
plotlib.setproperties(ax=cax, fontsize=15, yticklabelrot=0)
return ax
def plotBindingCurve(affinityData, variant, annotate=True, ax=None):
'''Plot a binding curve of a particular variant'''
# Load data
subSeries = affinityData.getVariantBindingSeries(variant)
if len(subSeries)==0:
print 'No fluorescence data associated with variant %s'%str(variant)
return
concentrations = affinityData.x
variant_table = affinityData.variant_table
# Plot
if ax is None:
fig = plt.figure(figsize=(4, 4))
ax = fig.add_subplot(111)
plotting.plotFitCurve(concentrations, subSeries, variant_table.loc[variant], ax=ax)
plotlib.setproperties(xlabel='Concentration (nM)', ylabel='Fluorescence',
labelfontsize=16, tickfontsize=16)
# Annonate
if annotate:
names = ['dG', 'dG_lb', 'dG_ub', 'fmax', 'fmax_lb', 'fmax_ub', 'numTests', 'pvalue', 'rsq', 'flag']
vec = pd.Series([processresults.getValueInTable(variant_table.loc[variant], name) for name in names], index=names)
#vec.fillna('', inplace=True)
annotationText = ['Kd = {:4.2f} uM ({:4.2f}, {:4.2f})'.format(liblib.dGtoKd(vec.dG, 'uM'),
liblib.dGtoKd(vec.dG_lb, 'uM'),
liblib.dGtoKd(vec.dG_ub, 'uM')),
'fmax = {:4.2f} ({:4.2f}, {:4.2f})'.format(vec.fmax,
vec.fmax_lb,
vec.fmax_ub),
'Nclusters = {:4.0f}'.format(vec.numTests),
'pvalue = {:.1e}'.format(vec.pvalue),
'average Rsq = {:4.2f}'.format(vec.rsq),
'fmax enforced = {}'.format(vec.flag)
]
ax.annotate('\n'.join(annotationText), xy=(0.05, .95), xycoords='axes fraction',
horizontalalignment='left', verticalalignment='top', fontsize=11)
return ax
def doubleMutant(allData_aggPF,
mutantSM,
targetSM,
normSM,
names,
norm=False,
coop=False):
# Define variables
data = allData_aggPF[targetSM]['sm_norm_diff_signals_2']['median']
mutantRefVariant = allData_aggPF[mutantSM]['sm_norm_diff_signals_2'][
'median'].idxmax()
targetRefVariant = allData_aggPF[targetSM]['sm_norm_diff_signals_2'][
'median'].idxmax()
normRefVariant = allData_aggPF[normSM]['sm_norm_diff_signals_2'][
'median'].idxmax()
if norm:
targetRefSignal = allData_aggPF[targetSM]['sm_norm_diff_signals_2'][
'median'][targetRefVariant]
normRefSignal = allData_aggPF[normSM]['sm_norm_diff_signals_2'][
'median'][normRefVariant]
else:
targetRefSignal = 1
normRefSignal = 1
libSeq = 'NNGGATTTTCCNNNNACGAAGTNNTCCCGAG'
startPos = 14
title = 'Mutants of ' + names[mutantSM].lower(
) + '0, in the presence of ' + names[targetSM]
if norm and coop:
cbarLabel = 'Cooperativity normalized to ' + names[targetSM].lower(
) + '0'
elif norm and not coop:
cbarLabel = 'Differential signals normalized to ' + names[
targetSM].lower() + '0'
elif not norm and coop:
cbarLabel = 'Cooperativity'
else:
cbarLabel = 'Differential signals'
# Compute vmin and vmax manually (assuming robust=True) so that the color bar
# scale is the same across the same dataRefVariant
normRefData = allData_aggPF[normSM]['sm_norm_diff_signals_2'][
'median'] / normRefSignal
doubleMutantSignals, mutantLabels = plotlib.doubleMutantMatrix(
normRefData, normRefVariant, libSeq, startPos, coop)
doubleMutantSignals = doubleMutantSignals[~np.isnan(doubleMutantSignals)]
vmin = np.percentile(doubleMutantSignals, 1)
vmax = np.percentile(doubleMutantSignals, 99)
vlim = max(abs(vmin), abs(vmax))
vmin, vmax = -vlim, vlim
# Make heatmap
plt.figure(figsize=(12, 10))
ax, cax = plotlib.doubleMutant(data,
mutantRefVariant,
libSeq,
startPos=startPos,
refSignal=targetRefSignal,
normToRefSignal=norm,
coop=coop,
vmin=vmin,
vmax=vmax,
cbarLabel=cbarLabel,
triangle='lower',
invertY=False,
linewidth=3)
cax.tick_params(labelsize=20)
cax.yaxis.label.set_fontsize(20)
plotlib.setproperties(title=title, labelfontsize=20, titlefontsize=20)
# Save to file
if coop:
figureOutput = 'coop'
else:
figureOutput = 'doubleMutant'
if norm:
figureNorm = 'norm_'
else:
figureNorm = ''
plt.savefig(figureDir + '/' + names[targetSM] + '_' +
names[mutantSM].lower() + '0_' + names[normSM].lower() + '0_' +
figureNorm + figureOutput + '.png')
return ax, cax