def generate_correction_range(self):
"""
Plot RT vs max(ri-rt) - min(ri-rt)
"""
diffs = pd.DataFrame(index=self.standards_dfs[0].index)
fig = Figure()
for sample in self.features_dfs:
name = self._get_name(sample)
diffs[name] = sample[(name, self.COLUMNS['RI'])] - sample[
(name, self.COLUMNS['RT'])]
diffs[self.COLUMNS['RT']] = self.standards_dfs[0][(
self.COLUMNS['TARGET'], self.COLUMNS['RT'])]
diffs.sort_values(by=self.COLUMNS['RT'])
fig.add_trace(
Scattergl(x=diffs[self.COLUMNS['RT']],
y=diffs.max() - diffs.min(),
mode='lines+markers',
**self.scatter_shared_attrs))
fig.update_layout(**self.standard_scatter_layout,
# height=3000,
)
self.__zones__(fig)
fig.update_xaxes(title=dict(text='Retention Time (s)', font_size=18),
showticklabels=True)
fig.update_yaxes(
title=dict(text='max(ri-rt) - min(ri-rt)', font_size=18))
fig.update_layout(
title_text='RT Correction error difference plot across samples')
return fig
def look_at_ss_rank_plot():
# make giant combination of all stocks and look at correlation between shortsqueeze rating 10 days and price change 10 days later
# no correlation, but this isn't super helpful, because it should be the
# ranking on the day new data comes out vs the price change 10 days later.
sh_int['A'].columns.tolist().index('Short Squeeze Ranking') # 145
all_feats, all_targs = [], []
for s in sh_int_stocks:
if sh_int[s].shape[0] > hist_points: # make_all_sh_future did this too
# print(s)
new_feats, new_targs, _, _ = create_hist_feats(
sh_int[s]['Short Squeeze Ranking'].values.reshape(-1, 1),
sh_int[s][targ_col].values,
sh_int[s].index.values,
hist_points=1,
future=10)
all_feats.append(new_feats)
all_targs.append(new_targs)
all_feats_np = np.concatenate(all_feats).flatten()
all_targs_np = np.concatenate(all_targs).flatten()
# no correlation, but this isn't super helpful, because it should be the
# ranking on the day new data comes out vs the price change 10 days later.
trace = Scattergl(x=all_feats_np,
y=all_targs_np,
mode='markers',
marker=dict(color='#FFBAD2', line=dict(width=1)))
data = [trace]
plot(data, filename='webgl')
def generate_istd_correlation(self):
"""Plots RI vs RT for ISTDs in each given sample"""
fig = Figure()
for sample in self.features_dfs:
sample_name = self._get_name(sample)
sample_color = next(self.sample_colors)
sample_rts = sample.loc[:, (sample_name, self.COLUMNS['RT'])]
sample_ris = sample.loc[:, (sample_name, self.COLUMNS['RI'])]
# regression calculation
# slope, intercept, r_value, p_value, std_err = stats.linregress(sample_rts.dropna(), sample_ris.dropna())
# line = slope * sample_rts.dropna().values + intercept
fig.add_trace(
Scattergl(x=sample_rts,
y=sample_ris,
name=sample_name,
hovertext=sample_rts.index,
legendgroup=sample_name,
line_color=sample_color,
mode='markers',
**self.scatter_shared_attrs))
# fig.add_trace(Scattergl(
# x=sample_rts,
# y=line,
# mode='line',
# name='Fit ' + sample_name,
# legendgroup=sample_name,
# line=dict(color=sample_color,
# dash='dash',
# width=1
# )
# ))
# fig.add_annotation(Annotation(
# x=3.5,
# y=23.5,
# text=f'$R^2 = {r_value},\\Y = {slope}*X + {intercept}$',
# showarrow=False,
# font=Font(size=16)
# ))
fig.update_layout(**self.scatter_layout,
legend=dict(x=1.0,
y=1.0,
xanchor='left',
yanchor='top',
orientation='v'))
fig.update_xaxes(showticklabels=True)
fig.update_yaxes(title=dict(text='Retention Index', font_size=18))
fig.update_layout(title=dict(
text='RT vs RI Correlation of correction ISTDs', font_size=18))
return fig
def create_traces(data, datetimes, n_channels, use_gradient):
logger.info("Creating traces")
traces = [
Scattergl(
x=datetimes,
y=data[:, i],
#xaxis=f'x{i + 1}',
yaxis=f'y{i + 1}',
line={
'color': ('rgb(0, 0, 255)'),
'width': 0.7
}) for i in range(n_channels)
]
if use_gradient:
for i, trace in enumerate(traces):
trace.line['color'] = \
f'rgb({255*i/n_channels}, 0, {255*(1-i/n_channels)})'
logger.info("Trace created")
return traces
def get_cluster_plot(species, grouping):
"""Generate tSNE cluster plot.
Arguments:
species (str): Name of species.
grouping (str): Which variable to use for grouping. cluster_ordered, biosample, layer or cluster_biosample
Returns:
str: HTML generated by Plot.ly.
"""
points_3d = get_3D_cluster_points(species)
if not points_3d:
points_2d = get_cluster_points(species)
if not points_2d:
raise FailToGraphException
if not (grouping in points_2d[0]):
grouping = "cluster_ordered"
print("**** Using cluster_ordered")
else:
if not (grouping in points_3d[0]):
grouping = "cluster_ordered"
print("**** Using cluster_ordered")
layout3d = Layout(autosize=False,
width=900,
height=700,
title='3D Cell Cluster',
titlefont={
'color': 'rgba(1,2,2,1)',
'size': 20
},
margin={
'l': 100,
'r': 150,
'b': 100,
't': 75
},
scene={
'camera': {
'eye': dict(x=1.2, y=1.5, z=0.7),
'center': dict(x=0.25, z=-0.1)
},
'aspectmode': 'data',
'xaxis': {
'title': 'tSNE 1',
'titlefont': {
'color': 'rgba(1,2,2,1)',
'size': 12
},
'type': 'linear',
'ticks': '',
'showticklabels': False,
'tickwidth': 0,
'showline': True,
'showgrid': False,
'zeroline': False,
'linecolor': 'black',
'linewidth': 0.5,
'mirror': True
},
'yaxis': {
'title': 'tSNE 2',
'titlefont': {
'color': 'rgba(1,2,2,1)',
'size': 12
},
'type': 'linear',
'ticks': '',
'showticklabels': False,
'tickwidth': 0,
'showline': True,
'showgrid': False,
'zeroline': False,
'linecolor': 'black',
'linewidth': 0.5,
'mirror': True
},
'zaxis': {
'title': 'tSNE 3',
'titlefont': {
'color': 'rgba(1,2,2,1)',
'size': 12
},
'type': 'linear',
'ticks': '',
'showticklabels': False,
'tickwidth': 0,
'showline': True,
'showgrid': False,
'zeroline': False,
'linecolor': 'black',
'linewidth': 0.5,
'mirror': True
}
})
layout2d = Layout(
autosize=True,
showlegend=True,
width=900,
height=700,
margin={
'l': 100,
'r': 150,
'b': 75,
't': 75,
'pad': 20
},
legend={
'orientation': 'v',
'traceorder': 'grouped',
'tracegroupgap': 10,
'x': 1.03,
'font': {
'color': 'rgba(1,2,2,1)',
'size': 12
},
},
xaxis={
'title': 'tSNE 1',
'titlefont': {
'color': 'rgba(1,2,2,1)',
'size': 16
},
'type': 'linear',
'ticks': '',
'showticklabels': False,
'tickwidth': 0,
'showline': True,
'showgrid': False,
'zeroline': False,
'linecolor': 'black',
'linewidth': 0.5,
'mirror': True
},
yaxis={
'title': 'tSNE 2',
'titlefont': {
'color': 'rgba(1,2,2,1)',
'size': 16
},
'type': 'linear',
'ticks': '',
'showticklabels': False,
'tickwidth': 0,
'showline': True,
'showgrid': False,
'zeroline': False,
'linecolor': 'black',
'linewidth': 0.5,
'mirror': True,
# 'range': [-20,20]
},
title='Cell clusters',
titlefont={
'color': 'rgba(1,2,2,1)',
'size': 20
},
annotations=[{
'text': 'Cluster',
'x': 0,
'y': -0.2,
'ax': 0,
'ay': 0,
'showarrow': False,
'font': {
'color': 'rgba(1,2,2,1)',
'size': 16
},
'xref': 'paper',
'yref': 'paper',
'xanchor': 'left',
'yanchor': 'bottom',
'textangle': 0,
}])
global trace_colors
trace_colors = dict()
traces_2d = OrderedDict()
global num_colors
if not points_3d:
max_cluster = int(
max(points_2d, key=lambda x: int(x['cluster_ordered']))
['cluster_ordered']) + 1
if species == 'mmu':
max_cluster = 16
num_colors = int(
max(points_2d, key=lambda x: int(x[grouping]))[grouping]) + 1
colors = generate_cluster_colors(num_colors)
symbols = [
'circle', 'square', 'cross', 'triangle-up', 'triangle-down',
'octagon', 'star', 'diamond'
]
for point in points_2d:
cluster_num = int(point['cluster_ordered'])
biosample = int(point.get('biosample', 1)) - 1
cluster_sample_num = int(
point['cluster_ordered']) + max_cluster * biosample
color_num = int(point[grouping]) - 1
trace2d = traces_2d.setdefault(
cluster_sample_num,
Scattergl(
x=list(),
y=list(),
text=list(),
mode='markers',
visible=True,
name=point['cluster_name'] + " Sample" +
str(biosample + 1),
legendgroup=point[grouping],
marker={
'color': colors[color_num],
'size': 7,
'symbol':
symbols[biosample], # Eran and Fangming 09/12/2017
'line': {
'width': 1,
'color': 'black'
}
},
hoverinfo='text'))
trace2d['x'].append(point['tsne_x'])
trace2d['y'].append(point['tsne_y'])
trace2d['text'].append(
build_hover_text({
'Cell': point.get('samp', 'N/A'),
'Layer': point.get('layer', 'N/A'),
'Biosample': point.get('biosample', 'N/A'),
'Cluster': str(cluster_num)
}))
for i, (key, value) in enumerate(sorted(traces_2d.items())):
trace_str = 'cluster_color_' + str(int(value['legendgroup']) - 1)
trace_colors.setdefault(trace_str, []).append(i)
if species == 'mmu':
for i in range(17, 23, 1):
traces_2d[i]['marker']['size'] = 15
traces_2d[i]['marker']['symbol'] = symbols[i % len(symbols)]
traces_2d[i]['marker']['color'] = 'black'
traces_2d[i]['visible'] = "legendonly"
return {'traces_2d': traces_2d, 'layout2d': layout2d}
else:
max_cluster = int(
max(points_3d, key=lambda x: int(x['cluster_ordered']))
['cluster_ordered']) + 1
if species == 'mmu':
max_cluster = 16
num_colors = int(
max(points_3d, key=lambda x: int(x[grouping]))[grouping]) + 1
colors = generate_cluster_colors(num_colors)
symbols = [
'circle', 'square', 'cross', 'triangle-up', 'triangle-down',
'octagon', 'star', 'diamond'
]
traces_3d = OrderedDict()
for point in points_3d:
cluster_num = int(point['cluster_ordered'])
biosample = int(point.get('biosample', 1)) - 1
cluster_sample_num = int(
point['cluster_ordered']) + max_cluster * biosample
color_num = int(point[grouping]) - 1
trace2d = traces_2d.setdefault(
cluster_sample_num,
Scattergl(
x=list(),
y=list(),
text=list(),
mode='markers',
visible=True,
name=point['cluster_name'] + " Sample" +
str(biosample + 1),
legendgroup=point[grouping],
marker={
'color': colors[color_num],
'size': 7,
'symbol':
symbols[biosample], # Eran and Fangming 09/12/2017
'line': {
'width': 1,
'color': 'black'
}
},
hoverinfo='text'))
trace2d['x'].append(point['tsne_x'])
trace2d['y'].append(point['tsne_y'])
trace2d['text'].append(
build_hover_text({
'Cell': point.get('samp', 'N/A'),
'Layer': point.get('layer', 'N/A'),
'Biosample': point.get('biosample', 'N/A'),
'Cluster': str(cluster_num)
}))
trace3d = traces_3d.setdefault(
cluster_sample_num,
Scatter3d(
x=list(),
y=list(),
z=list(),
text=list(),
mode='markers',
visible=True,
name=point['cluster_name'] + " Sample" +
str(biosample + 1),
legendgroup=point[grouping],
marker={
'size': 4,
'symbol':
symbols[biosample], # Eran and Fangming 09/12/2017
'line': {
'width': 1,
'color': 'black'
},
'color': colors[color_num],
},
hoverinfo='text'))
trace3d['x'].append(point['tsne_1'])
trace3d['y'].append(point['tsne_2'])
trace3d['z'].append(point['tsne_3'])
trace3d['text'] = trace2d['text']
for i, (key, value) in enumerate(sorted(traces_2d.items())):
trace_str = 'cluster_color_' + str(int(value['legendgroup']) - 1)
trace_colors.setdefault(trace_str, []).append(i)
if species == 'mmu':
for i in range(17, 23, 1):
traces_2d[i]['marker']['size'] = traces_3d[i]['marker'][
'size'] = 15
traces_2d[i]['marker']['symbol'] = traces_3d[i]['marker'][
'symbol'] = symbols[i % len(symbols)]
traces_2d[i]['marker']['color'] = traces_3d[i]['marker'][
'color'] = 'black'
traces_2d[i]['visible'] = traces_3d[i][
'visible'] = "legendonly"
return {
'traces_2d': traces_2d,
'traces_3d': traces_3d,
'layout2d': layout2d,
'layout3d': layout3d
}
def alignment_plot_eic(self):
"""
Returns a Figure representing the extracted ion chromatogram for the given samples. Plots rt and ri vs
intensity per standard
"""
masses = dict(self.standards_dfs[0].loc[:, (self.COLUMNS['TARGET'],
self.COLUMNS['MASS'])])
rts = dict(
self.standards_dfs[0].loc[:, (self.COLUMNS['TARGET'],
self.COLUMNS['RT'])].sort_values())
fig = make_subplots(rows=2,
cols=1,
vertical_spacing=0.15,
shared_xaxes=True,
shared_yaxes=True)
self.max_x = 0
self.max_y = 0
# chromatogram plot
for rawdata_df in self.rawdata_dfs:
sample_name = self._get_name(rawdata_df)
rawdata_df = rawdata_df.xs(sample_name, axis=1, drop_level=True)
rawdata_summed_rt = pd.DataFrame(index=rawdata_df.index)
rawdata_summed_ri = pd.DataFrame(index=rawdata_df.index)
for tgt, rt in rts.items():
target_color = self.target_colors[tgt]
# rt v intensity
rawdata_summed_rt = rawdata_df[abs(rawdata_df[self.COLUMNS['PRECMZ']] - masses[tgt] <= self.mass_tolerance)] \
.groupby(self.COLUMNS['RT'])[self.COLUMNS['INTENSITY']].sum()
fig.add_trace(
Scattergl(
x=rawdata_df.groupby(
self.COLUMNS['RT'])[self.COLUMNS['RT']].mean(),
y=rawdata_summed_rt.transpose(),
legendgroup=tgt,
marker_color=target_color,
name=f"{rt:.2f} {re.sub(self.INCHIKEY_REGEX, '', tgt)}",
mode='lines',
**self.scatter_shared_attrs),
col=1,
row=1,
)
# ri v intensity
rawdata_summed_ri = rawdata_df[abs(rawdata_df[self.COLUMNS['PRECMZ']] - masses[tgt] <= self.mass_tolerance)] \
.groupby(self.COLUMNS['RI'])[self.COLUMNS['INTENSITY']].sum()
fig.add_trace(
Scattergl(x=rawdata_df.groupby(
self.COLUMNS['RT'])[self.COLUMNS['RT']].mean(),
y=rawdata_summed_ri.transpose(),
legendgroup=tgt,
marker_color=target_color,
showlegend=False,
mode='lines',
**self.scatter_shared_attrs),
col=1,
row=2,
)
self.max_x = max(self.max_x, rawdata_df[self.COLUMNS['RI']].max(),
rawdata_df[self.COLUMNS['RT']].max())
self.max_y = max(self.max_y, max(rawdata_summed_ri),
max(rawdata_summed_rt))
fig.update_layout(paper_bgcolor='white',
**self.standard_scatter_layout)
fig.update_yaxes(title_text='Intensity',
tickformat='e',
range=[0, self.max_y],
showticklabels=True)
fig.update_xaxes(range=[0, self.max_x], showticklabels=True)
fig.update_layout(title_text="ISTD alignment plot (EIC)")
return fig
def plot_results(mod,
tr_feats,
tr_targs,
te_feats=None,
te_targs=None,
folder=None,
test_lin_preds=None):
# TODO: subplot with actual and predicted returns
train_preds = mod.predict(tr_feats)[:, 0]
train_score = mod.evaluate(tr_feats, tr_targs)
# couldn't figure this out yet
# train_score = K.run(stock_loss_mae_log(train_targs, train_preds))
title = 'train preds vs actual (score = ' + str(train_score) + ')'
if te_feats is None:
title = 'full train preds vs actual (score = ' + str(train_score) + ')'
data = [
Scattergl(x=train_preds,
y=tr_targs,
mode='markers',
name='preds vs actual',
marker=dict(color=list(range(tr_targs.shape[0])),
colorscale='Portland',
showscale=True,
opacity=0.5))
]
layout = Layout(title=title,
xaxis=dict(title='predictions'),
yaxis=dict(title='actual'))
fig = Figure(data=data, layout=layout)
if folder is None:
if te_feats is None:
filename = plot_dir + 'full_train_preds_vs_actual.html'
else:
filename = plot_dir + 'train_preds_vs_actual.html'
else:
if not os.path.exists(plot_dir + folder):
os.mkdir(plot_dir + folder)
if te_feats is None:
filename = plot_dir + folder + '/' + 'full_train_preds_vs_actual.html'
else:
filename = plot_dir + folder + '/' + 'train_preds_vs_actual.html'
plot(fig, filename=filename, auto_open=False, show_link=False)
del train_score
if te_feats is not None:
test_preds = mod.predict(te_feats)[:, 0]
test_score = mod.evaluate(te_feats, te_targs)
# test_score = K.run(stock_loss_mae_log(train_targs, train_preds))
data = [
Scattergl(x=test_preds,
y=te_targs,
mode='markers',
name='preds vs actual',
marker=dict(color=list(range(te_targs.shape[0])),
colorscale='Portland',
showscale=True,
opacity=0.5))
]
if test_lin_preds is not None:
line = Scatter(x=test_preds,
y=test_lin_preds,
mode='lines',
name='linear fit')
data = data + [line]
layout = Layout(title='test preds vs actual (score = ' +
str(test_score) + ')',
xaxis=dict(title='predictions'),
yaxis=dict(title='actual'))
fig = Figure(data=data, layout=layout)
if folder is None:
filename = plot_dir + 'test_preds_vs_actual.html'
else:
filename = plot_dir + folder + '/' + 'test_preds_vs_actual.html'
plot(fig, filename=filename, auto_open=False, show_link=False)
del test_score
from planet import Planet
import plotly.offline as plt
from plotly.graph_objs import Scattergl
earth = Planet("La tierra", 0.017, 1, 365.26, 5.9742 * 10**24, 0, 0, 101.22)
orbit = earth.get_orbit_2bodies(200)
plt.plot({'data': [Scattergl(x=orbit[:, 0], y=orbit[:, 1])]})
def generate_rt_shift(self) -> Figure:
"""
Creates a chromatogram plot showing the sift produced by retention time correction
Returns:
fig: A plotly.graph_objs.Figure consisting of two subplots: rt vs intensity on top and ri vs intensity
on the bottom.
Raw data is plotted as a bar chart and standards are overlaid as a scatter plot.
"""
fig = Figure()
height = 100 + 50 * (len(self.rawdata_dfs) // 7)
# chromatogram plot
for rawdata_df in self.rawdata_dfs:
sample_color = next(self.sample_colors)
sample_name = rawdata_df.columns.unique(level=0).values[0]
rawdata_df = rawdata_df.xs(sample_name, axis=1, drop_level=True)
# rt v intensity
rawdata_summed_rt = rawdata_df.groupby(self.COLUMNS['RT'])[
self.COLUMNS['INTENSITY']].sum().reset_index()
fig.add_trace(
Scattergl(x=rawdata_summed_rt[self.COLUMNS['RT']],
y=rawdata_summed_rt[self.COLUMNS['INTENSITY']],
legendgroup=sample_name,
marker_color=sample_color,
name='RT ' + sample_name,
mode='lines',
**self.scatter_shared_attrs))
# ri v intensity
rawdata_summed_ri = rawdata_df.groupby(self.COLUMNS['RI'])[
self.COLUMNS['INTENSITY']].sum().reset_index()
fig.add_trace(
Scattergl(x=rawdata_summed_ri[self.COLUMNS['RI']],
y=rawdata_summed_ri[self.COLUMNS['INTENSITY']],
legendgroup=sample_name,
marker_color=sample_color,
line={'dash': 'dash'},
name='RI ' + sample_name,
mode='lines',
**self.scatter_shared_attrs))
self.max_x = max(self.max_x, rawdata_df[self.COLUMNS['RI']].max(),
rawdata_df[self.COLUMNS['RT']].max())
self.max_y = max(
self.max_y, rawdata_summed_ri[self.COLUMNS['INTENSITY']].max(),
rawdata_summed_rt[self.COLUMNS['INTENSITY']].max())
# fig = self._add_zones_and_std_lines(fig)
fig.update_layout(**self.scatter_layout,
xaxis_showticklabels=True,
height=height,
legend=dict(x=0.5,
y=-0.1,
xanchor='center',
yanchor='top',
orientation='h'))
fig.update_xaxes(title_text='RT (solid) - RI (dashed)',
range=[0, self.max_x * 1.1])
fig.update_yaxes(title_text='Intensity', tickformat='e')
fig.update_layout(title_text="Retention Time correction comparison")
return fig
def alignment_plot_tic(self):
"""Plots total ion chromatogram for given samples"""
# FIXME Fix this plot. Must have to do with the x axis
fig = make_subplots(rows=2,
cols=1,
vertical_spacing=0.15,
shared_xaxes=True,
shared_yaxes=True)
max_x = 0
max_y = 0
height = 1000 + (50 * (len(self.rawdata_dfs) // 7))
# chromatogram plot
for rawdata_df in self.rawdata_dfs:
sample_color = next(self.sample_colors)
sample_name = rawdata_df.columns.unique(level=0).values[0]
rawdata_df = rawdata_df.xs(sample_name, axis=1, drop_level=True)
line_type = next(self.line_types)
# rt v intensity
rawdata_summed_rt = rawdata_df.groupby(self.COLUMNS['RT'])[
self.COLUMNS['INTENSITY']].sum().reset_index()
fig.add_trace(
Scattergl(x=rawdata_summed_rt[self.COLUMNS['RT']],
y=rawdata_summed_rt[self.COLUMNS['INTENSITY']],
legendgroup=sample_name,
line=dict(width=1),
marker_color=sample_color,
name='RT ' + sample_name,
mode='lines',
**self.scatter_shared_attrs),
col=1,
row=1,
)
# ri v intensity
rawdata_summed_ri = rawdata_df.groupby(self.COLUMNS['RI'])[
self.COLUMNS['INTENSITY']].sum().reset_index()
fig.add_trace(
Scattergl(x=rawdata_summed_ri[self.COLUMNS['RI']],
y=rawdata_summed_ri[self.COLUMNS['INTENSITY']],
legendgroup=sample_name,
line=dict(dash=line_type, width=1),
marker_color=sample_color,
name='RI ' + sample_name,
mode='lines',
**self.scatter_shared_attrs),
col=1,
row=2,
)
max_x = max(max_x, rawdata_df[self.COLUMNS['RI']].max(),
rawdata_df[self.COLUMNS['RT']].max())
max_y = max(max_y,
rawdata_summed_ri[self.COLUMNS['INTENSITY']].max(),
rawdata_summed_rt[self.COLUMNS['INTENSITY']].max())
fig.update_layout(**self.scatter_layout,
paper_bgcolor='white',
height=height,
legend=dict(x=0.5,
y=-0.1,
xanchor='center',
yanchor='top',
orientation='h'))
fig.update_xaxes(title_text='RT (solid) - RI (dashed)',
range=[0, max_x * 1.1],
showticklabels=True)
fig.update_yaxes(title_text='Intensity',
tickformat='e',
range=[max_y * -0.05, max_y * 1.1])
fig.update_layout(title_text="Retention Time correction comparison")
return fig
def generate_rt_ri_delta_from_reference(self):
"""
Creates two plots showing the difference in retention time (top) and retention index (bottom) versus the
library reference for each internal standard
"""
fig = make_subplots(rows=2,
cols=1,
vertical_spacing=0.05,
shared_xaxes=True)
deltas_dfs = []
for s in self.standards_dfs:
s = s.sort_values([(self.COLUMNS['TARGET'], self.COLUMNS['RT'])],
axis='index')
sample_name = self._get_name(s)
deltas_dfs.append(self._calc_diff(s, sample_name))
min_y = 0
max_y = 0
for sample in deltas_dfs:
sample_name = self._get_name(sample)
sample_color = next(self.sample_colors)
size = len(sample.index.values)
sample.columns = sample.columns.droplevel(0)
min_y = min(min_y, sample['rt_diff'].min())
max_y = max(max_y, sample['rt_diff'].max())
min_y = min(min_y, sample['ri_diff'].min())
max_y = max(max_y, sample['ri_diff'].max())
min_y = min(min_y, sample['corr_diff'].min())
max_y = max(max_y, sample['corr_diff'].max())
fig.add_trace(
Scattergl(x=self._fix_names(sample.index.values),
y=[0] * size,
name='RT ' + sample_name,
legendgroup=sample_name,
line_color=sample_color,
error_y=dict(
type='data',
symmetric=False,
array=sample['rt_diff'],
),
**self.scatter_shared_attrs),
col=1,
row=1,
)
fig.add_trace(
Scattergl(x=self._fix_names(sample.index.values),
y=[0] * size,
name='RI ' + sample_name,
legendgroup=sample_name,
line_color=sample_color,
error_y=dict(
type='data',
symmetric=False,
array=sample['ri_diff'],
),
**self.scatter_shared_attrs),
col=1,
row=2,
)
fig.add_hline(y=self.ri_tolerance,
line=dict(color='rgba(0,0,0,0.75)',
dash='dot',
width=1))
fig.add_hline(y=-self.ri_tolerance,
line=dict(color='rgba(0,0,0,0.75)',
dash='dot',
width=1))
fig.update_layout(
**self.standard_scatter_layout,
height=3000,
# paper_bgcolor='white',
)
fig.update_xaxes(type='category')
fig['layout']['xaxis'].update(showticklabels=False)
fig['layout']['xaxis2'].update(
showticklabels=True,
tickangle=60,
title=dict(text='ISTD Name (Sorted L-R by increasing RT)',
font_size=18))
fig['layout']['yaxis'].update(title=dict(
text='Lib vs sample RT Difference', font_size=18),
range=[min_y * 1.1, max_y * 1.1])
fig['layout']['yaxis2'].update(title=dict(
text='Lib vs sample RI Difference', font_size=18),
range=[min_y * 1.1, max_y * 1.1])
fig.update_layout(
title_text=
'Retention Time and Index deviations from the Reference (library)'
)
return fig
def generate_plot(self) -> Figure:
"""
Creates a chromatogram plot showing the sift produced by retention time correction
"""
fig = Figure()
height = 1000 + (len(self.rawdata_dfs) * 50 // 7)
# chromatogram plot
for rawdata_df in self.rawdata_dfs:
sample_color = next(
self.sample_colors
) # TODO Have this cycle reset every time generate_plot is called
sample_name = rawdata_df.columns.unique(level=0).values[0]
rawdata_df = rawdata_df.xs(sample_name, axis=1, drop_level=True)
line_type = next(self.line_types)
# rt v intensity
rawdata_summed_rt = rawdata_df.groupby(self.COLUMNS['RT'])[
self.COLUMNS['INTENSITY']].sum().reset_index()
fig.add_trace(
Scattergl(
x=rawdata_summed_rt[self.COLUMNS['RT']],
y=rawdata_summed_rt[self.COLUMNS['INTENSITY']],
opacity=self.rt_line2_opacity,
# showlegend=False,
legendgroup=sample_name,
line=dict(width=1), # dash=line_type,
marker_color=sample_color,
name='RT ' + sample_name,
mode='lines',
**self.scatter_shared_attrs))
# ri v intensity
rawdata_summed_ri = rawdata_df.groupby(self.COLUMNS['RI'])[
self.COLUMNS['INTENSITY']].sum().reset_index()
fig.add_trace(
Scattergl(x=rawdata_summed_ri[self.COLUMNS['RI']],
y=rawdata_summed_ri[self.COLUMNS['INTENSITY']],
opacity=self.ri_line2_opacity,
legendgroup=sample_name,
marker_color=sample_color,
mode='lines',
line=dict(dash='dash', width=1),
name='RI ' + sample_name,
**self.scatter_shared_attrs))
self.max_x = max(self.max_x, rawdata_df[self.COLUMNS['RI']].max(),
rawdata_df[self.COLUMNS['RT']].max())
self.max_y = max(
self.max_y, rawdata_summed_ri[self.COLUMNS['INTENSITY']].max(),
rawdata_summed_rt[self.COLUMNS['INTENSITY']].max())
# markers for found ISTDs -- x = RT, + = RI
for standards_df in self.standards_dfs:
sample_name = standards_df.columns.unique(level=0).values[1]
marker_color = next(self.sample_colors)
standards_int = standards_df.loc[:, (sample_name,
self.COLUMNS['INTENSITY'])]
annot_rt = standards_df.loc[:, (sample_name, self.COLUMNS['RT'])]
annot_ri = standards_df.loc[:, (sample_name, self.COLUMNS['RI'])]
# error_bars
error_bars = self._calc_diff(standards_df, sample_name)
rt_error_bars = error_bars.droplevel(
level=0, axis=1)[self.COLUMNS['RT_DIFF']]
ri_error_bars = error_bars.droplevel(
level=0, axis=1)[self.COLUMNS['RI_DIFF']]
fig.add_trace(
Scattergl(x=annot_rt,
y=standards_int,
showlegend=False,
legendgroup=sample_name,
text=standards_df.index,
mode='markers',
name=sample_name,
marker=dict(size=10,
symbol='x',
opacity=self.rt_marker_opacity,
color=marker_color,
line=dict(color=marker_color, width=1)),
error_x=dict(
array=rt_error_bars,
symmetric=False,
),
**self.scatter_shared_attrs))
fig.add_trace(
Scattergl(x=annot_ri,
y=standards_int,
showlegend=False,
legendgroup=sample_name,
text=standards_df.index,
line=dict(dash='dash', width=1),
mode='markers',
name=sample_name,
marker=dict(size=10,
symbol='cross',
opacity=self.ri_marker_opacity,
color=marker_color,
line=dict(color=marker_color, width=1)),
error_x=dict(
array=ri_error_bars,
symmetric=False,
),
**self.scatter_shared_attrs))
fig = self._add_zones_and_std_lines(fig)
fig.update_layout(paper_bgcolor='white',
xaxis_showticklabels=True,
height=height,
**self.standard_scatter_layout)
#Override parent legend
fig.update_layout(legend=dict(
x=0.5, y=-0.1, xanchor='center', yanchor='top', orientation='h'))
fig.update_layout(
title_text="Annotated ISTDs before and after RT Correction")
fig.update_xaxes(title_text='RT (solid) - RI (dashed)',
range=[0, self.max_x * 1.1])
fig.update_yaxes(title_text='Intensity',
tickformat='e',
range=[self.max_y * -0.05, self.max_y * 1.1])
return fig