30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232 | def spatial_interaction (adata,
spatial_interaction='spatial_interaction',
summarize_plot=True,
p_val=0.05,
row_cluster=False,
col_cluster=False,
cmap = 'vlag',
nonsig_color='grey',
subset_phenotype=None,
subset_neighbour_phenotype=None,
binary_view=False,
return_data=False, **kwargs):
"""
Parameters:
adata (anndata.AnnData):
The annotated data matrix with spatial interaction calculations.
spatial_interaction (str, optional):
Key in `adata.uns` where spatial interaction data is stored, typically the output of `sm.tl.spatial_interaction`.
summarize_plot (bool, optional):
If True, summarizes cell-cell interactions across all images or samples to provide an aggregated view.
p_val (float, optional):
Threshold for significance of interactions. Interactions with a P-value above this threshold are considered non-significant.
row_cluster, col_cluster (bool, optional):
If True, performs hierarchical clustering on rows or columns in the heatmap to group similar patterns of interaction.
cmap (str, optional):
Colormap for the heatmap visualization. Default is 'vlag'.
nonsig_color (str, optional):
Color used to represent non-significant interactions in the heatmap.
subset_phenotype, subset_neighbour_phenotype (list, optional):
Subsets of phenotypes or neighboring phenotypes to include in the analysis and visualization.
binary_view (bool, optional):
If True, visualizes interactions in a binary manner, highlighting presence or absence of significant interactions without intensity gradation.
return_data (bool, optional):
If True, returns the DataFrame used for plotting instead of the plot itself.
**kwargs:
Additional keyword arguments for seaborn's clustermap function, such as `linecolor` and `linewidths`.
Returns:
pandas.DataFrame (dataframe):
Only if `return_data` is True. The DataFrame containing the data used for plotting.
Example:
```python
# Basic visualization of spatial interactions with default settings
sm.pl.spatial_interaction(adata)
# Detailed heatmap of spatial interactions, excluding non-significant interactions
sm.pl.spatial_interaction(adata, summarize_plot=False, p_val=0.01, cmap='coolwarm', nonsig_color='lightgrey',
binary_view=True, row_cluster=True, col_cluster=True)
# Visualizing specific phenotypes interactions, with custom colormap and binary view
sm.pl.spatial_interaction(adata, subset_phenotype=['T cells', 'B cells'], subset_neighbour_phenotype=['Macrophages'],
cmap='seismic', binary_view=True, row_cluster=True, col_cluster=False,
figsize=(10, 8), dendrogram_ratio=(.1, .2), cbar_pos=(0, .2, .03, .4))
```
"""
# set color for heatmap
#cmap_updated = copy.copy(matplotlib.cm.get_cmap(cmap))
#cmap_updated = matplotlib.cm.get_cmap(cmap)
cmap_updated = matplotlib.colormaps[cmap]
cmap_updated.set_bad(color=nonsig_color)
# Copy the interaction results from anndata object
try:
interaction_map = adata.uns[spatial_interaction].copy()
except KeyError:
raise ValueError('spatial_interaction not found- Please run sm.tl.spatial_interaction first')
# subset the data if user requests
if subset_phenotype is not None:
if isinstance(subset_phenotype, str):
subset_phenotype = [subset_phenotype]
# subset the phenotype
interaction_map = interaction_map[interaction_map['phenotype'].isin(subset_phenotype)]
if subset_neighbour_phenotype is not None:
if isinstance(subset_neighbour_phenotype, str):
subset_neighbour_phenotype = [subset_neighbour_phenotype]
# subset the phenotype
interaction_map = interaction_map[interaction_map['neighbour_phenotype'].isin(subset_neighbour_phenotype)]
# Seperate Interaction intensity from P-value
p_value = interaction_map.filter(regex='pvalue_')
p_val_df = pd.concat([interaction_map[['phenotype','neighbour_phenotype']], p_value], axis=1, join='outer')
p_val_df = p_val_df.set_index(['phenotype','neighbour_phenotype'])
interaction_map = interaction_map[interaction_map.columns.difference(p_value.columns)]
interaction_map = interaction_map.set_index(['phenotype','neighbour_phenotype'])
# Binarize the values if user requests
if binary_view == True:
interaction_map[interaction_map > 0] = 1
interaction_map[interaction_map <= 0] = -1
if summarize_plot == True:
# convert first two columns to multi-index column
#interaction_map = interaction_map.set_index(['phenotype','neighbour_phenotype'])
#p_val_df = p_val_df.set_index(['phenotype','neighbour_phenotype'])
# If multiple images are present, take the average of interactions
interaction_map['mean'] = interaction_map.mean(axis=1).values
interaction_map = interaction_map[['mean']] # keep only the mean column
interaction_map = interaction_map['mean'].unstack()
# Do the same for P-values
p_val_df['mean'] = p_val_df.mean(axis=1).values
p_val_df = p_val_df[['mean']] # keep only the mean column
# set the P-value threshold
p_val_df.loc[p_val_df[p_val_df['mean'] > p_val].index,'mean'] = np.NaN
p_val_df = p_val_df['mean'].unstack()
# change to the order passed in subset
if subset_phenotype is not None:
interaction_map = interaction_map.reindex(subset_phenotype)
p_val_df = p_val_df.reindex(subset_phenotype)
if subset_neighbour_phenotype is not None:
interaction_map = interaction_map.reindex(columns=subset_neighbour_phenotype)
p_val_df = p_val_df.reindex(columns=subset_neighbour_phenotype)
# Plotting heatmap
mask = p_val_df.isnull() # identify the NAN's for masking
im = interaction_map.fillna(0) # replace nan's with 0 so that clustering will work
# heatmap
sns.clustermap(im, cmap=cmap_updated, row_cluster=row_cluster, col_cluster=col_cluster, mask=mask, **kwargs)
else:
if len(interaction_map.columns) < 2:
raise ValueError('Data for only a single image is available please set summarize_plot=True and try again')
# convert first two columns to multi-index column
#interaction_map = interaction_map.set_index(['phenotype','neighbour_phenotype'])
#p_val_df = p_val_df.set_index(['phenotype','neighbour_phenotype'])
# P value threshold
p_val_df = p_val_df.apply(lambda x: np.where(x > p_val,np.nan,x))
# Remove rows that are all nan
idx = p_val_df.index[p_val_df.isnull().all(1)] # Find all nan rows
interaction_map = interaction_map.loc[interaction_map.index.difference(idx)] # clean intensity data
p_val_df = p_val_df.loc[p_val_df.index.difference(idx)] # clean p-value data
# order the plot as needed
if subset_phenotype or subset_neighbour_phenotype is not None:
interaction_map.reset_index(inplace=True)
p_val_df.reset_index(inplace=True)
if subset_phenotype is not None:
interaction_map['phenotype'] = interaction_map['phenotype'].astype('str').astype('category')
interaction_map['phenotype'] = interaction_map['phenotype'].cat.reorder_categories(subset_phenotype)
interaction_map = interaction_map.sort_values('phenotype')
# Do same for Pval
p_val_df['phenotype'] = p_val_df['phenotype'].astype('str').astype('category')
p_val_df['phenotype'] = p_val_df['phenotype'].cat.reorder_categories(subset_phenotype)
p_val_df = p_val_df.sort_values('phenotype')
if subset_neighbour_phenotype is not None:
interaction_map['neighbour_phenotype'] = interaction_map['neighbour_phenotype'].astype('str').astype('category')
interaction_map['neighbour_phenotype'] = interaction_map['neighbour_phenotype'].cat.reorder_categories(subset_neighbour_phenotype)
interaction_map = interaction_map.sort_values('neighbour_phenotype')
# Do same for Pval
p_val_df['neighbour_phenotype'] = p_val_df['neighbour_phenotype'].astype('str').astype('category')
p_val_df['neighbour_phenotype'] = p_val_df['neighbour_phenotype'].cat.reorder_categories(subset_neighbour_phenotype)
p_val_df = p_val_df.sort_values('neighbour_phenotype')
if subset_phenotype and subset_neighbour_phenotype is not None:
interaction_map = interaction_map.sort_values(['phenotype', 'neighbour_phenotype'])
p_val_df = p_val_df.sort_values(['phenotype', 'neighbour_phenotype'])
# convert the data back into multi-index
interaction_map = interaction_map.set_index(['phenotype', 'neighbour_phenotype'])
p_val_df = p_val_df.set_index(['phenotype', 'neighbour_phenotype'])
# Plotting heatmap
mask = p_val_df.isnull() # identify the NAN's for masking
im = interaction_map.fillna(0) # replace nan's with 0 so that clustering will work
mask.columns = im.columns
# covert the first two columns into index
# Plot
sns.clustermap(im, cmap=cmap_updated, row_cluster=row_cluster, col_cluster=col_cluster, mask=mask, **kwargs)
if return_data is True:
# perpare data for export
map_data = interaction_map.copy()
p_val_data = mask.copy()
map_data.reset_index(inplace=True)
p_val_data.reset_index(inplace=True)
# remove the first two colums
map_data = map_data.drop(['phenotype','neighbour_phenotype'],axis=1)
p_val_data = p_val_data.drop(['phenotype','neighbour_phenotype'],axis=1)
p_val_data.columns = map_data.columns
# remove the mased values
final_Data = map_data.where(~p_val_data, other=np.nan)
final_Data.index = interaction_map.index
return final_Data
|