Dimensionality FiguresΒΆ
In [31]:
import sys
import os
from sciviso import *
from scproximite import *
Set-up¶
Load data and create visualisation Dataframe¶
In [32]:
data_dir = '../results/'
all_datasets = ['Discrete_Abundant','Discrete_Abundant_HVG2000','Discrete_Abundant_HVG500',
'Discrete_Rare','Discrete_Rare_HVG2000','Discrete_Rare_HVG500',
'Continuous_Abundant','Continuous_Abundant_HVG2000','Continuous_Abundant_HVG500',
'Continuous_Rare','Continuous_Rare_HVG2000','Continuous_Rare_HVG500']
datasets = dict.fromkeys(all_datasets, None)
In [33]:
all_metrics = ['euclidean', 'cityblock', 'correlation', 'spearman', 'cosine', 'braycurtis', 'hamming', 'jaccard',
'kulsinski', 'yule', 'OT', 'phi_s', 'weighted_rank', 'kendall', 'zi_kendall','canberra', 'chebyshev']
eval_metrics=['PSI']
k_vals = ["3K_stats", '10K_stats', "30K_stats", "50K_stats"]
for dataset in datasets.keys():
datasets[dataset] = load_pickle(dataset, data_dir, mode='mean', metrics= all_metrics,eval_metrics=eval_metrics,k_vals=k_vals)
In [34]:
vis_df = vis_dataframe(datasets=all_datasets, datasets_dict=datasets, metrics=all_metrics, k_vals=k_vals, eval_metrics=eval_metrics)
#Create visualisation dataframe for supplementary Figure 11 (heatmap):
k30_df = vis_dataframe(datasets=all_datasets, datasets_dict=datasets, metrics=all_metrics, k_vals=['30K_stats'], eval_metrics=eval_metrics)
In [35]:
#Get the mean PSI for each metric, grouped by dataset across all k values:
mean_df = vis_df.groupby(['Metric','Dataset'])['Performance'].describe()
mean_df['ID'] = mean_df.index.map(lambda x: '_'.join([str(i) for i in x]))
dat = mean_df.reset_index()
dat = dat[['mean', 'ID']]
vis_df['ID'] = vis_df['Metric'].str.cat(vis_df[['Dataset']], sep='_')
vis_df = vis_df.merge(dat, how='inner', on='ID')
In [36]:
delta_df = vis_df.drop(labels = ['Performance','k_value','Evaluation_Metric',
'Population_Balance'], axis = 1)
delta_df = delta_df.drop_duplicates(keep='first')
#Calculate difference in PSI between the dimensionality levels:
for index, row in delta_df.iterrows():
met = row['Metric']
con = row['Condition']
Base = delta_df.query(f"Metric == '{met}' and Property=='Base' and Condition=='{con}'")
HVG2000 = delta_df.query(f"Metric == '{met}' and Property=='HVG2000' and Condition=='{con}'")
HVG500 = delta_df.query(f"Metric == '{met}' and Property=='HVG500' and Condition=='{con}'")
delta_df.loc[index,'Base'] = Base['mean'].values
delta_df.loc[index,'HVG2000'] = HVG2000['mean'].values
delta_df.loc[index,'HVG500'] = HVG500['mean'].values
delta_df.loc[index,'Base to HVG500'] = HVG500['mean'].values - Base['mean'].values
delta_df.loc[index,'Base to HVG2000'] = HVG2000['mean'].values - Base['mean'].values
delta_df.loc[index,'HVG2000 to HVG500'] = HVG500['mean'].values - HVG2000['mean'].values
Visualisation Aesthetics¶
In [37]:
figure_dir = ''
plt.rcParams['svg.fonttype'] = 'none'
plt.rcParams['figure.dpi'] = 300
# Set to true if you want to save figures out:
save = False
#heatmap settings
heatmap_config = {
'title_font_size': 15,
'label_font_size': 12,
'axis_font_size': 10,
's': 60,
'figsize': (15, 10),
}
#Seaborn theme and context:
sns.set_context("notebook")
sns.set_style("darkgrid",{'ytick.left': True,'xtick.bottom': True, 'axes.linewidth':1, 'axes.edgecolor':'black'})
#Colour dictionary for Metrics:
colour_dict = {'euclidean':'#393b79', 'cityblock':'#6b6ecf', 'correlation':'#637939', 'spearman':'#b5cf6b', 'cosine':'#e6550d',
'braycurtis':'#843c39', 'hamming':'#d6616b', 'jaccard':'#31a354', 'kulsinski':'#7b4173', 'yule':'#ce6dbd',
'OT':'#e7ba52', 'phi_s':'#3182bd','weighted_rank':'#e7969c', 'kendall':'#fdae6b', 'zi_kendall':'#9467bd',
'canberra':'#8c6d31', 'chebyshev':'#636363'}
#Colour scale for heatmaps:
base_scale = 'viridis'
Figure 5¶
In [38]:
fig5A_df = delta_df.copy(deep=True)
In [39]:
#Filter to only include metrics which show a >= change in PSI for one or both of the reduced dimension datasets relative to the HD.
for index, row in fig5A_df.iterrows():
if row['Base to HVG500'] >=0.050 or row['Base to HVG2000'] >=0.050:
fig5A_df.loc[index,'pass'] = True
else:
fig5A_df.loc[index,'pass'] = False
fig5A_df = fig5A_df[fig5A_df['pass'] == True]
In [40]:
import warnings
# Ignore setting with copy warning
warnings.simplefilter(action='ignore')
#Split vis dataframe by structural condition (e.g. Discrete Abundant):
adatas = [fig5A_df[fig5A_df['Condition'].isin([conditions])] for conditions in fig5A_df['Condition'].unique()]
new_adatas = []
for data in adatas:
for index, row in data.iterrows():
if row['Base to HVG500'] > row['Base to HVG2000']:
data.loc[index,'Change'] = row['Base to HVG500']
else:
data.loc[index,'Change'] = row['Base to HVG2000']
data = data.sort_values(by=['Change'], ascending=False) #sort dataframe from largest change in PSI to smallest
data.reset_index(drop=True,inplace=True) #reset the index so it matches the new sort
data = data[0:15] #Subset to only retain the top 5 metrics (accounting for duplicate rows for the 3 dimensionality levels).
new_adatas.append(data)
In [41]:
filtered_data = pd.concat(new_adatas)
#Sort dataframe to ensure individual figure legends are correct:
filtered_data['Condition'] = pd.Categorical(filtered_data['Condition'], categories=['Discrete Abundant','Discrete Rare', 'Continuous Abundant', 'Continuous Rare'], ordered=True)
filtered_data = filtered_data.sort_values(['Metric','Condition']).reset_index(drop=True)
In [42]:
data = filtered_data
g = sns.catplot(x = 'Property',
y = 'mean',
data = data,
hue = 'Metric',
style='Metric',
col = 'Condition',
col_order = ['Discrete Abundant','Discrete Rare', 'Continuous Abundant', 'Continuous Rare'],
col_wrap = 2,
palette = colour_dict,
s=10,
kind='point',
join=True,
order = ['Base', 'HVG2000', 'HVG500'],
dodge=0.06,
aspect = 1.8,
sharey =True,
height = 2.5,
scale = 0.6,
ci=None,
legend_out = False).despine(left=True)
#### To produce individual legends for each subplot: ######
ser_vals = pd.Series(data['Metric'].sort_values().unique())
for axes, (i, d) in zip(g.axes.ravel(), data.groupby(['Condition'])):
handles, labels = axes.get_legend_handles_labels()
#subplot aesthetics
axes.margins(x=0.05)
#axes.yaxis.set_tick_params(which='both', labelbottom=True)
axes.xaxis.set_tick_params(which='both', labelbottom=True)
vals = ser_vals[ser_vals.isin(d['Metric'].unique())]
idx = vals.index.tolist()
if len(idx) > 0:
axes.legend(handles = [handles[i] for i in idx], labels = vals.tolist(),bbox_to_anchor=(1, 1),
loc=2,prop={'size': 7},markerscale=1.3,facecolor='white',handletextpad=0,frameon=False)
############################################################
g.fig.subplots_adjust(wspace=0.3,hspace=0.45)
g.set_xlabels('Dimensionality Condition', fontsize=11)
g.set_ylabels('Clustering Performance (PSI)',fontsize=10)
g.set_titles(template='{col_name}')
if save is True:
plt.savefig(f'{fig_dir}Figure_5A.svg', bbox_inches = 'tight')
plt.savefig(f'{fig_dir}Figure_5A.png', bbox_inches = 'tight', dpi=300)
Figure 5B¶
In [43]:
fig5B_df = delta_df.copy(deep=True)
In [44]:
#Filter to only include metrics which show an absolute change in PSI between the reduced dimension datasets of > 0.05:
for index, row in fig5B_df.iterrows():
if abs(round(row['HVG2000 to HVG500'],2))>0.05:
fig5B_df.loc[index,'pass'] = True
else:
fig5B_df.loc[index,'pass'] = False
fig5B_df = fig5B_df[fig5B_df['pass'] == True]
In [45]:
fig5B_df['Condition'] = pd.Categorical(fig5B_df['Condition'], categories=['Discrete Abundant','Discrete Rare', 'Continuous Abundant', 'Continuous Rare'], ordered=True)
fig5B_df = fig5B_df.sort_values(['Metric','Condition']).reset_index(drop=True)
In [46]:
data = fig5B_df
g = sns.catplot(x = 'Property',
y = 'mean',
data = data,
hue = 'Metric',
style='Metric',
col = 'Condition',
col_order = ['Discrete Abundant','Discrete Rare', 'Continuous Abundant', 'Continuous Rare'],
col_wrap = 2,
palette = colour_dict,
s=10,
kind='point',
join=True,
order = ['Base', 'HVG2000', 'HVG500'],
dodge=0.06,
aspect = 1.65,
sharey = True,
height = 2.43,
scale = 0.6,
ci=None,
legend_out = False).despine(left=True)
ser_vals = pd.Series(data['Metric'].sort_values().unique())
for axes, (i, d) in zip(g.axes.ravel(), data.groupby(['Condition'])):
handles, labels = axes.get_legend_handles_labels()
axes.margins(x=0.05)
#axes.yaxis.set_tick_params(which='both', labelbottom=True)
axes.xaxis.set_tick_params(which='both', labelbottom=True)
vals = ser_vals[ser_vals.isin(d['Metric'].unique())]
idx = vals.index.tolist()
if len(idx) > 0:
axes.legend(handles = [handles[i] for i in idx], labels = vals.tolist(), loc=2,
bbox_to_anchor=(1, 1),prop={'size': 9}, markerscale=1.3,
facecolor='white',handletextpad=0, frameon=False)
g.fig.subplots_adjust(left=0.11, bottom = 0.15, wspace=0.4,hspace=0.45)
g.set_ylabels('')
g.set_xlabels(' ')
g.set_titles(template='{col_name}')
g.fig.supxlabel('Dimensionality Condition',fontsize=15)
g.fig.supylabel('Clustering Performance (PSI)', fontsize=15)
g.set(ylim=(0, 1))
if save is True:
plt.savefig(f'{fig_dir}Figure_5B.svg', bbox_inches = 'tight')
plt.savefig(f'{fig_dir}Figure_5B.png', bbox_inches = 'tight', dpi=300)
Supplementary Figure 4¶
In [47]:
vis_df['Condition'] = pd.Categorical(vis_df['Condition'], categories=['Discrete Abundant','Discrete Rare', 'Continuous Abundant', 'Continuous Rare'], ordered=True)
vis_df = vis_df.sort_values(['Metric','Condition']).reset_index(drop=True)
In [48]:
data = vis_df
g = sns.catplot(x = 'Property',
y = 'mean',
data = data,
hue = 'Metric',
style='Metric',
col = 'Condition',
col_order = ['Discrete Abundant','Discrete Rare', 'Continuous Abundant', 'Continuous Rare'],
col_wrap = 2,
palette = colour_dict,
s=10,
kind='point',
join=True,
order = ['Base', 'HVG2000', 'HVG500'],
dodge=0.06,
aspect = 1.7,
sharey = True,
height = 2.5,
scale = 0.6,
ci=None,
legend=False,
legend_out = False).despine(left=True)
for axes, (i, d) in zip(g.axes.ravel(), data.groupby(['Condition'])):
handles, labels = axes.get_legend_handles_labels()
axes.margins(x=0.05)
#axes.yaxis.set_tick_params(which='both', labelbottom=True)
axes.xaxis.set_tick_params(which='both', labelbottom=True)
g.fig.legend(handles = handles, bbox_to_anchor=(1.15,0.8),
prop={'size': 10},markerscale=1.3,facecolor='white',handletextpad=0,frameon=False)
g.fig.subplots_adjust(left=0.1,wspace=0.1,hspace=0.35)
g.set_ylabels('')
g.set_xlabels(' ')
g.set_titles(template='{col_name}')
g.fig.supxlabel('Dimensionality Condition',fontsize=15)
g.fig.supylabel('Clustering Performance (PSI)', fontsize=15)
if save is True:
plt.savefig(f'{fig_dir}Figure_Sup4.svg', bbox_inches = 'tight')
plt.savefig(f'{fig_dir}Figure_Sup4.png', bbox_inches = 'tight', dpi=300)
Supplementary Figure 5¶
In [49]:
figSup5_df = delta_df.copy(deep=True)
In [50]:
#Filter to only include metrics which show an absolute change in PSI between the reduced dimension datasets of > 0.1:
for index, row in figSup5_df.iterrows():
if row['Base to HVG500'] >0.10 or row['Base to HVG2000'] >0.10:
figSup5_df.loc[index,'pass'] = True
else:
figSup5_df.loc[index,'pass'] = False
figSup5_df = figSup5_df[figSup5_df['pass'] == True]
In [51]:
figSup5_df['Condition'] = pd.Categorical(figSup5_df['Condition'], categories=['Discrete Abundant','Discrete Rare', 'Continuous Abundant', 'Continuous Rare'], ordered=True)
figSup5_df = figSup5_df.sort_values(['Metric','Condition']).reset_index(drop=True)
In [52]:
data = figSup5_df
g = sns.catplot(x = 'Property',
y = 'mean',
data = data,
hue = 'Metric',
style='Metric',
col = 'Condition',
col_order = ['Discrete Abundant','Discrete Rare', 'Continuous Abundant', 'Continuous Rare'],
col_wrap = 2,
palette = colour_dict,
s=10,
kind='point',
join=True,
order = ['Base', 'HVG2000', 'HVG500'],
dodge=0.06,
aspect = 1.7,
sharey = True,
height = 2.5,
scale = 0.6,
ci=None,
legend_out = False).despine(left=True)
ser_vals = pd.Series(data['Metric'].sort_values().unique())
for axes, (i, d) in zip(g.axes.ravel(), data.groupby(['Condition'])):
handles, labels = axes.get_legend_handles_labels()
axes.margins(x=0.05)
#axes.yaxis.set_tick_params(which='both', labelbottom=True)
axes.xaxis.set_tick_params(which='both', labelbottom=True)
vals = ser_vals[ser_vals.isin(d['Metric'].unique())]
idx = vals.index.tolist()
if len(idx) > 0:
axes.legend(handles = [handles[i] for i in idx], labels = vals.tolist(), loc=2,
bbox_to_anchor=(1,1),prop={'size': 9}, markerscale=1.3,
facecolor='white',handletextpad=0, frameon=False)
g.fig.subplots_adjust(left=0.1, bottom = 0.15, wspace=0.4,hspace=0.45)
g.set_ylabels('')
g.set_xlabels(' ')
g.set_titles(template='{col_name}')
g.fig.supxlabel('Dimensionality Condition',fontsize=15)
g.fig.supylabel('Clustering Performance (PSI)', fontsize=15)
if save is True:
plt.savefig(f'{fig_dir}Figure_Sup5.svg', bbox_inches = 'tight')
plt.savefig(f'{fig_dir}Figure_Sup5.png', bbox_inches = 'tight', dpi=300)
Supplementary Figure 6¶
In [53]:
figSup6_df = delta_df.copy(deep=True)
In [54]:
#Filter to only include metrics which have <0.05 change in PSI between all dimensionalities:
for index, row in figSup6_df.iterrows():
if abs(row['Base to HVG500'])<0.050 and abs(row['Base to HVG2000'])<0.050 and abs(row['HVG2000 to HVG500'])<0.050:
figSup6_df.loc[index,'pass'] = True
else:
figSup6_df.loc[index,'pass'] = False
figSup6_df = figSup6_df[figSup6_df['pass'] == True]
In [55]:
figSup6_df['Condition'] = pd.Categorical(figSup6_df['Condition'], categories=['Discrete Abundant','Discrete Rare', 'Continuous Abundant', 'Continuous Rare'], ordered=True)
figSup6_df = figSup6_df.sort_values(['Metric','Condition']).reset_index(drop=True)
In [56]:
data = figSup6_df
g = sns.catplot(x = 'Property',
y = 'mean',
data = data,
hue = 'Metric',
style='Metric',
col = 'Condition',
col_order = ['Discrete Abundant','Discrete Rare', 'Continuous Abundant', 'Continuous Rare'],
col_wrap = 2,
palette = colour_dict,
s=10,
kind='point',
join=True,
order = ['Base', 'HVG2000', 'HVG500'],
dodge=0.06,
aspect = 1.7,
sharey = True,
height = 2.35,
scale = 0.6,
ci=None,
legend_out = False).despine(left=True)
ser_vals = pd.Series(data['Metric'].sort_values().unique())
for axes, (i, d) in zip(g.axes.ravel(), data.groupby(['Condition'])):
handles, labels = axes.get_legend_handles_labels()
axes.margins(x=0.05)
#axes.yaxis.set_tick_params(which='both', labelbottom=True)
axes.xaxis.set_tick_params(which='both', labelbottom=True)
vals = ser_vals[ser_vals.isin(d['Metric'].unique())]
idx = vals.index.tolist()
if len(idx) > 0:
axes.legend(handles = [handles[i] for i in idx], labels = vals.tolist(),loc=2,
bbox_to_anchor=(0.95, 1),prop={'size': 9}, markerscale=1.3,
facecolor='white',handletextpad=0, frameon = False)
g.fig.subplots_adjust(left=0.11, bottom = 0.15, wspace=0.45,hspace=0.45)
g.set_ylabels('')
g.set_xlabels(' ')
g.set_titles(template='{col_name}')
g.fig.supxlabel('Dimensionality Condition',fontsize=15)
g.fig.supylabel('Clustering Performance (PSI)', fontsize=15)
if save is True:
plt.savefig(f'{fig_dir}Figure_Sup6.png', bbox_inches = 'tight', dpi=300)
plt.savefig(f'{fig_dir}Figure_Sup6.svg', bbox_inches = 'tight')
Heatmap Figures¶
Figure 10 Panel B¶
In [57]:
#Drop unnecessary columns & remove duplicates:
heat_df = vis_df.drop(labels = ['Performance','k_value','Evaluation_Metric',
'Population_Balance', 'Data_Structure', 'Condition', 'ID', 'Property'], axis = 1)
heat_df = heat_df.drop_duplicates(keep='first').set_index('Metric')
#Reform dataframe:
heat_df = heat_df.pivot(columns='Dataset', values='mean').reindex(columns=all_datasets)
#Sort metrics by mean best performing across all datasets:
heat_df['metric_avg'] = np.mean(heat_df.values, axis=1)
heat_df = heat_df.sort_values(by=['metric_avg'], ascending=False)
heat_df['metric'] = heat_df.index
heatmap = Heatmap(heat_df,chart_columns=all_datasets,
row_index='metric',
cmap=base_scale,
vmin=0, vmax=1,
annot = True,
cluster_rows=False, cluster_cols=False,
config=heatmap_config)
ax = heatmap.plot()
ax.ax_heatmap.tick_params(axis='x', which='major', pad=10)
if save is True:
plt.savefig(f'{fig_dir}Figure_10B.svg', bbox_inches = 'tight')
plt.savefig(f'{fig_dir}Figure10B.png', bbox_inches = 'tight', dpi=300)
Supplementary Figure 11 Panel B¶
In [58]:
heat_df = k30_df.drop(labels = ['Evaluation_Metric', 'Data_Structure', 'Population_Balance', 'Condition', 'Property', 'k_value'], axis = 1).set_index('Metric')
heat_df = heat_df.pivot(columns='Dataset', values='Performance').reindex(columns=all_datasets)
heat_df['metric_avg'] = np.mean(heat_df.values, axis=1)
heat_df = heat_df.sort_values(by=['metric_avg'], ascending=False)
heat_df['metric'] = heat_df.index
heatmap = Heatmap(heat_df,chart_columns=all_datasets,
row_index='metric',
cmap=base_scale,
vmin=0, vmax=1,
annot = True,
cluster_rows=False, cluster_cols=False,
config=heatmap_config)
ax = heatmap.plot()
ax.ax_heatmap.tick_params(axis='x', which='major', pad=10)
if save is True:
plt.savefig(f'{fig_dir}Figure_Sup11B.png', bbox_inches = 'tight')
plt.show()
In [ ]: