MESA Ecospatial Tutorials (hepatocellular carcinoma, CosMx)
[1]:
import numpy as np
import pandas as pd
import anndata as ad
import seaborn as sns
from scipy import stats
import matplotlib.pyplot as plt
import matplotlib.colors as mcolors
from matplotlib.figure import figaspect
import os
os.sys.path.append('../../../')
from mesa import ecospatial as eco
/opt/miniconda3/envs/mesa/lib/python3.11/site-packages/geopandas/_compat.py:106: UserWarning: The Shapely GEOS version (3.8.0-CAPI-1.13.1) is incompatible with the GEOS version PyGEOS was compiled with (3.10.4-CAPI-1.16.2). Conversions between both will be slow.
warnings.warn(
OMP: Info #276: omp_set_nested routine deprecated, please use omp_set_max_active_levels instead.
/opt/miniconda3/envs/mesa/lib/python3.11/site-packages/spaghetti/network.py:41: FutureWarning: The next major release of pysal/spaghetti (2.0.0) will drop support for all ``libpysal.cg`` geometries. This change is a first step in refactoring ``spaghetti`` that is expected to result in dramatically reduced runtimes for network instantiation and operations. Users currently requiring network and point pattern input as ``libpysal.cg`` geometries should prepare for this simply by converting to ``shapely`` geometries.
warnings.warn(dep_msg, FutureWarning, stacklevel=1)
[2]:
plt.rcParams['font.family'] = 'Arial'
plt.rcParams['svg.fonttype'] = 'none' # To keep text as editable text in SVGs
Helper functions
[3]:
import time
from typing import Union,List
def overlap_check(new_patch, existing_patches, max_overlap_ratio):
"""
This function checks if the new patch overlaps with any of the existing patches.
Parameters:
new_patch: tuple
The coordinates of the new patch to be checked for overlaps.
existing_patches: list
A list of existing patches to check for overlaps.
max_overlap_ratio: float
The maximum allowable overlap ratio for a new patch.
Returns:
bool
Returns True if the new patch doesn't overlap with any existing patch beyond the allowable overlap ratio.
Otherwise, returns False.
"""
patch_area = (new_patch[2] - new_patch[0]) * (new_patch[3] - new_patch[1])
max_overlap = max_overlap_ratio * patch_area
for patch in existing_patches:
dx = min(new_patch[2], patch[2]) - max(new_patch[0], patch[0])
dy = min(new_patch[3], patch[3]) - max(new_patch[1], patch[1])
if (dx>=0) and (dy>=0) and (dx*dy > max_overlap):
return False
return True
def contains_points(patch, spatial_values, min_points):
"""
This function checks if a patch contains a certain number of spatial values (points).
Parameters:
patch: tuple
The coordinates of the patch to be checked.
spatial_values: list
A list of spatial values (cells' coordinates) to check if they are within the patch.
min_points: int
The minimum number of points that the patch should contain.
Returns:
bool
Returns True if the patch contains at least 'min_points' number of spatial values（cells' coordinates）. Otherwise, returns False.
"""
# Count the points within the patch
points_in_patch = sum((patch[0] <= point[0] <= patch[2]) and (patch[1] <= point[1] <= patch[3]) for point in spatial_values)
# Check if the number of points in the patch is at least min_points
return points_in_patch >= min_points
def subsampling_randomly(spatial_data: Union[ad.AnnData, pd.DataFrame],
library_key:str,
library_id:str,
sample_size:Union[int,float],
sample_number:int,
spatial_key:Union[str,List[str]],
max_overlap=0.0,
random_seed=None,
min_points=2):
rng = np.random.default_rng(random_seed)
if isinstance(spatial_data, ad.AnnData):
spatial_data_filtered = spatial_data[spatial_data.obs[library_key] == library_id]
elif isinstance(spatial_data, pd.DataFrame):
spatial_data_filtered = spatial_data[spatial_data[library_key] == library_id]
else:
raise ValueError("spatial_data should be either an AnnData object or a pandas DataFrame")
if isinstance(spatial_data, ad.AnnData):
spatial_values = spatial_data_filtered.obsm[spatial_key]
elif isinstance(spatial_data, pd.DataFrame):
spatial_values = spatial_data_filtered[spatial_key].values
else:
raise ValueError("spatial_data should be either an AnnData object or a pandas DataFrame")
width = spatial_values.max(axis=0)[0] - spatial_values.min(axis=0)[0]
height = spatial_values.max(axis=0)[1] - spatial_values.min(axis=0)[1]
patch_width = sample_size[0]
patch_height = sample_size[1] if sample_size[1] else sample_size[0]
num_patches = sample_number
patches = []
used_coordinates = set()
for _ in range(num_patches):
start_time = time.time()
while True:
if time.time() - start_time > 5: # seconds
print(f"Warning: Could not generate a new patch within 5 seconds. Return {len(patches)} out of {num_patches} patches")
return patches
x0 = rng.uniform(spatial_values.min(axis=0)[0], spatial_values.max(axis=0)[0] - patch_width)
y0 = rng.uniform(spatial_values.min(axis=0)[1], spatial_values.max(axis=0)[1] - patch_height)
x1 = x0 + patch_width
y1 = y0 + patch_height
new_patch = (x0, y0, x1, y1)
if ((x0, y0) not in used_coordinates and
overlap_check(new_patch, patches, max_overlap) and
contains_points(new_patch, spatial_values, min_points) ):
used_coordinates.add((x0, y0))
patches.append(new_patch)
break
return patches
Read Data
[4]:
liver_df = pd.read_csv('/Users/Emrys/Dropbox/spatial_augmentation/data/cosMx_liver/data/cosMx_liver_meta.csv')
[5]:
liver_df = liver_df[['cell_ID', 'Run_Tissue_name', 'x_slide_mm', 'y_slide_mm', 'fov', 'nCell', 'cellType', 'niche']].copy()
liver_df['x_slide_um'] = (liver_df['x_slide_mm']*1000.0).round(0)
liver_df['y_slide_um'] = (liver_df['y_slide_mm']*1000.0).round(0)
liver_df.head()
[5]:
| cell_ID | Run_Tissue_name | x_slide_mm | y_slide_mm | fov | nCell | cellType | niche | x_slide_um | y_slide_um | |
|---|---|---|---|---|---|---|---|---|---|---|
| 0 | c_1_100_10 | NormalLiver | 9.03144 | 9.73500 | 100 | 1265 | Hep.3 | Zone_3a | 9031.0 | 9735.0 |
| 1 | c_1_100_1078 | NormalLiver | 8.77440 | 9.25824 | 100 | 1265 | Hep.4 | Zone_2a | 8774.0 | 9258.0 |
| 2 | c_1_100_1135 | NormalLiver | 8.87928 | 9.23412 | 100 | 1265 | Inflammatory.macrophages | Zone_2a | 8879.0 | 9234.0 |
| 3 | c_1_100_267 | NormalLiver | 9.12132 | 9.61104 | 100 | 1265 | Hep.5 | Zone_2a | 9121.0 | 9611.0 |
| 4 | c_1_100_732 | NormalLiver | 9.08436 | 9.40548 | 100 | 1265 | Central.venous.LSECs | Zone_2a | 9084.0 | 9405.0 |
[6]:
liver_df.loc[:, 'sample'] = '-1' # Initialize the 'patches' column with -1
patches_coord_dict = {}
for tissue_type in pd.unique(liver_df['Run_Tissue_name']):
print(f"Processing {tissue_type}", flush=True)
# patches_coord = subsampling(liver_df,
# 'Run_Tissue_name',
# tissue_type,
# num_squares=[11.0,8.0],
# spatial_key=['x_slide_um', 'y_slide_um'])
patches_coord = subsampling_randomly(spatial_data=liver_df,
library_key='Run_Tissue_name',
library_id=tissue_type,
sample_size=[1600,1600],
sample_number=10,
spatial_key=['x_slide_um', 'y_slide_um'],
max_overlap=0.0,
random_seed=42,
min_points=12000)
patches_coord_dict[tissue_type] = patches_coord
print(f"Total number of patches: {len(patches_coord)}",flush=True)
print(patches_coord)
# Iterate over patches
for i, patch in enumerate(patches_coord):
x0, y0, x1, y1 = patch
# Define the number of decimal places for rounding
decimal_places = 5
# When assigning patch coordinates
x0 = round(x0, decimal_places)
y0 = round(y0, decimal_places)
x1 = round(x1, decimal_places)
y1 =round(y1, decimal_places)
# Update the 'patches' column for cells within the current patch
if tissue_type == 'NormalLiver':
sample_name = 'N'+str(i)
else:
sample_name = 'C'+str(i)
liver_df.loc[
(liver_df['Run_Tissue_name'] == tissue_type) &
(liver_df['x_slide_um'] >= x0) &
(liver_df['x_slide_um'] <= x1) &
(liver_df['y_slide_um'] >= y0) &
(liver_df['y_slide_um'] <= y1),
'sample'
] = sample_name
liver_df.drop(liver_df[ (liver_df['sample'] == '-1')].index, inplace=True)
Processing NormalLiver
Total number of patches: 10
[(1886.6148806917008, 10025.667690297836, 3486.614880691701, 11625.667690297836), (7967.310663046048, 8786.71629929017, 9567.310663046048, 10386.71629929017), (2196.0119891029517, 6592.722490085427, 3796.0119891029517, 8192.722490085427), (4408.565586928444, 9706.33596711446, 6008.565586928444, 11306.33596711446), (5070.607250708777, 5134.232285585302, 6670.607250708777, 6734.232285585302), (2302.5292319298533, 4397.568560631123, 3902.5292319298533, 5997.568560631123), (7121.82980264373, 6728.084803351514, 8721.82980264373, 8328.084803351514), (2416.6294404359564, 8200.149971508083, 4016.6294404359564, 9800.149971508083), (9767.184347122256, 6801.335263858796, 11367.184347122256, 8401.335263858797), (4050.9145487720484, 7052.114822553971, 5650.914548772049, 8652.114822553971)]
Processing CancerousLiver
Total number of patches: 10
[(8733.940845498759, 6956.522134005294, 10333.940845498759, 8556.522134005294), (959.7836280173957, 9924.939247260912, 2559.7836280173956, 11524.939247260912), (7711.752316668252, 3299.0830280094356, 9311.752316668251, 4899.083028009436), (7918.160165240158, 1593.40573536479, 9518.160165240159, 3193.40573536479), (1571.1241343269614, 6803.766233190506, 3171.124134326961, 8403.766233190505), (4669.1734372249175, 5584.331078425471, 6269.1734372249175, 7184.331078425471), (1642.1339927717302, 4862.145660802682, 3242.13399277173, 6462.145660802682), (2705.8975080433142, 9856.173593545518, 4305.897508043314, 11456.173593545518), (6011.6105483660085, -239.7283038551871, 7611.6105483660085, 1360.271696144813), (9435.311143736182, 3758.3641717244664, 11035.311143736182, 5358.364171724466)]
Perform Ecospatial Analysis on one sample
[43]:
test_sample = 'N7'
patches_coordinates = eco.generate_patches(liver_df,
'sample',
test_sample,
scaling_factor=32.,
spatial_key=['x_slide_um', 'y_slide_um'])
patch_indices, patches_comp = eco.calculate_diversity_index(spatial_data=liver_df,
library_key='sample',
library_id=test_sample,
spatial_key=['x_slide_um', 'y_slide_um'],
patches=patches_coordinates,
cluster_key='cellType',
metric='Shannon Diversity', return_comp=True)
# Visualize the heterogeneity indices
grid, heatmap_fig = eco.diversity_heatmap(spatial_data=liver_df,
library_key='sample',
library_id=test_sample,
spatial_key=['x_slide_um', 'y_slide_um'],
patches=patches_coordinates,
heterogeneity_indices=patch_indices,
tissue_only=True,
plot=True,
return_fig=True)
1.953 per cent patches are empty
[44]:
print(eco.global_spatial_stats(grid, tissue_only=True))
Global spatial stats restricted to tissue region with w of shape (1004, 1004)
(0.09882209548413516, 0.001)
[45]:
hotspots, coldspots = eco.local_spatial_stats(grid,tissue_only=True,p_value=0.01,plot_weights=False)
print(sum(hotspots.flatten()), flush=True)
print(sum(coldspots.flatten()), flush=True)
(1004,)
Local spatial stats restricted to tissue region with w of shape(1004, 1004)
Using MoranI
26
21
[41]:
Hot=True
if Hot:
filtered_patches_coordinates = [patch for patch, is_hotspot in zip(patches_coordinates, hotspots.flatten()) if is_hotspot]
filtered_patches_comp = [patch for patch, is_hotspot in zip(patches_comp, hotspots.flatten()) if is_hotspot]
else:
filtered_patches_coordinates = [patch for patch, is_coldspot in zip(patches_coordinates, coldspots.flatten()) if is_coldspot]
filtered_patches_comp = [patch for patch, is_coldspot in zip(patches_comp, coldspots.flatten()) if is_coldspot]
[42]:
pd.concat(filtered_patches_comp, axis=1).sum(axis=1)
[42]:
cellType
tumor_2 42.0
tumor_1 290.0
Periportal.LSECs 31.0
Stellate.cells 91.0
CD3+.alpha.beta.T.cells 184.0
Inflammatory.macrophages 206.0
Non.inflammatory.macrophages 53.0
NK.like.cells 6.0
Central.venous.LSECs 14.0
Mature.B.cells 39.0
gamma.delta.T.cells.1 3.0
Portal.endothelial.cells 2.0
Cholangiocytes 1.0
dtype: float64
Perform Ecospatial Analysis on all samples
[7]:
library_ids = liver_df['sample'].unique().tolist()
library_ids = [
l for l in library_ids if 'N' in l or 'C' in l
]
Calculate Global Cell Frequency & Cell Co-Occurrence
[8]:
global_cellfreq_df, global_co_occurrence_df = eco.spot_cellfreq(spatial_data=liver_df,
scale=32.0,
library_key='sample',
library_id=library_ids,
spatial_key=['x_slide_um', 'y_slide_um'],
cluster_key='cellType',
spots='global',
top=None,
selected_comb=None,
restricted=False,
metric='Shannon Diversity')
Processing region: N1 at scale 32.0
1.465 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: N7 at scale 32.0
1.953 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: N9 at scale 32.0
0.000 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: N6 at scale 32.0
0.195 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: N8 at scale 32.0
0.195 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: N2 at scale 32.0
0.000 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: N4 at scale 32.0
0.000 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: N5 at scale 32.0
0.098 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: N0 at scale 32.0
0.684 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: N3 at scale 32.0
0.195 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: C4 at scale 32.0
0.000 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: C0 at scale 32.0
0.000 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: C1 at scale 32.0
1.953 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: C5 at scale 32.0
0.000 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: C6 at scale 32.0
4.297 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: C7 at scale 32.0
0.000 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: C9 at scale 32.0
1.953 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: C2 at scale 32.0
0.000 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: C3 at scale 32.0
2.930 per cent patches are empty
Using MoranI
Considering whole tissue
Processing region: C8 at scale 32.0
6.348 per cent patches are empty
Using MoranI
Considering whole tissue
[9]:
global_cellfreq_df['Condition'] = ' '
global_cellfreq_df.loc[global_cellfreq_df.index.str.contains('N'), 'Condition'] = 'Healthy'
global_cellfreq_df.loc[global_cellfreq_df.index.str.contains('C'), 'Condition'] = 'HCC'
global_co_occurrence_subcols = global_co_occurrence_df.loc[:,global_co_occurrence_df.mean()>0.05].columns.tolist()
global_co_occurrence_df['Condition'] = ' '
global_co_occurrence_df['sample'] = global_co_occurrence_df.index
global_co_occurrence_df.loc[global_co_occurrence_df.index.str.contains('N'), 'Condition'] = 'Healthy'
global_co_occurrence_df.loc[global_co_occurrence_df.index.str.contains('C'), 'Condition'] = 'HCC'
global_co_occurrence_subcols.extend([('Condition',''),('sample','')])
[10]:
# Melt the dataframe for easier plotting and statistical analysis
global_cellfreq_df_melt = global_cellfreq_df.reset_index().melt(id_vars=['sample', 'Condition'])
global_cellfreq_df_melt.columns = ['sample', 'group', 'cellType', 'Frequency']
global_cellfreq_df_melt
[10]:
| sample | group | cellType | Frequency | |
|---|---|---|---|---|
| 0 | C0 | HCC | Antibody.secreting.B.cells | 0.001204 |
| 1 | C1 | HCC | Antibody.secreting.B.cells | 0.000233 |
| 2 | C2 | HCC | Antibody.secreting.B.cells | 0.003202 |
| 3 | C3 | HCC | Antibody.secreting.B.cells | 0.004519 |
| 4 | C4 | HCC | Antibody.secreting.B.cells | 0.001152 |
| ... | ... | ... | ... | ... |
| 435 | N5 | Healthy | tumor_2 | 0.000000 |
| 436 | N6 | Healthy | tumor_2 | 0.000000 |
| 437 | N7 | Healthy | tumor_2 | 0.000000 |
| 438 | N8 | Healthy | tumor_2 | 0.000000 |
| 439 | N9 | Healthy | tumor_2 | 0.000000 |
440 rows × 4 columns
[11]:
# Perform t-tests
selected_cell_types = sorted(liver_df['cellType'].unique())
selected_p_values = []
for ct in selected_cell_types:
group1 = global_cellfreq_df_melt[(global_cellfreq_df_melt['cellType'] == ct) & (global_cellfreq_df_melt['group'] == 'Healthy')]['Frequency']
group2 = global_cellfreq_df_melt[(global_cellfreq_df_melt['cellType'] == ct) & (global_cellfreq_df_melt['group'] == 'HCC')]['Frequency']
t_stat, p_value = stats.ttest_ind(group1, group2, equal_var=False)
print(f"{ct} has p value of {p_value}")
selected_p_values.append(p_value)
pvals_corrected = stats.false_discovery_control(selected_p_values, method='bh')
print('-'*42)
print(f"p-values after correction:")
# Plot
fig, ax = plt.subplots(figsize=(30,10))
sns.boxplot(data=global_cellfreq_df_melt, x='cellType', y='Frequency', hue='group', palette='muted', boxprops=dict(alpha=.3), ax=ax, dodge=True, order=selected_cell_types)
sns.swarmplot(data=global_cellfreq_df_melt, x='cellType', y='Frequency', hue='group', palette='dark:black', size=2.0, dodge=True, order=selected_cell_types, ax=ax, edgecolor='auto', linewidth=0.5)
handles, labels = ax.get_legend_handles_labels()
ax.legend(handles[:2], labels[:2], title="Groups", handletextpad=1, columnspacing=1, bbox_to_anchor=(1, 1), ncol=3, frameon=True)
plt.xticks(rotation=90)
p_vals_corrected_dict = {}
yrange = ax.get_ylim()[1] - ax.get_ylim()[0]
for i, ct in enumerate(selected_cell_types):
ax.text(i, yrange, f"p = {pvals_corrected[i]:.3f}", ha='center', fontsize=12, rotation=0)
print(f"{ct} has p value = {pvals_corrected[i]:.3f}", flush=True)
p_vals_corrected_dict[ct] = pvals_corrected[i]
for i in range(len(selected_cell_types) - 1):
ax.axvline(i + 0.55, color='grey', linestyle='--', linewidth=0.5)
ax.set_ylabel("Frequency", fontsize=14)
ax.set_xlabel('')
plt.show()
Antibody.secreting.B.cells has p value of 0.7253105069107637
CD3+.alpha.beta.T.cells has p value of 0.6688961489467853
Central.venous.LSECs has p value of 1.3650725054199515e-09
Cholangiocytes has p value of 5.184474315672871e-05
Erthyroid.cells has p value of 1.962577398118726e-08
Hep has p value of 0.3128233608731258
Hep.1 has p value of 1.4655635733085782e-05
Hep.3 has p value of 2.517630979080062e-10
Hep.4 has p value of 5.087963668099555e-10
Hep.5 has p value of 3.9066463036613865e-11
Hep.6 has p value of 2.0930984873598823e-06
Inflammatory.macrophages has p value of 0.14611237799276483
Mature.B.cells has p value of 0.5028132228052369
NK.like.cells has p value of 3.7138952557959555e-07
Non.inflammatory.macrophages has p value of 0.3171705903870426
NotDet has p value of 0.9868237859133563
Periportal.LSECs has p value of 0.3792861739107447
Portal.endothelial.cells has p value of 0.042454116173829735
Stellate.cells has p value of 0.0038202017195151047
gamma.delta.T.cells.1 has p value of 8.353966741526583e-06
tumor_1 has p value of 6.801927672934746e-07
tumor_2 has p value of 0.002071201681775486
------------------------------------------
p-values after correction:
Antibody.secreting.B.cells has p value = 0.760
CD3+.alpha.beta.T.cells has p value = 0.736
Central.venous.LSECs has p value = 0.000
Cholangiocytes has p value = 0.000
Erthyroid.cells has p value = 0.000
Hep has p value = 0.410
Hep.1 has p value = 0.000
Hep.3 has p value = 0.000
Hep.4 has p value = 0.000
Hep.5 has p value = 0.000
Hep.6 has p value = 0.000
Inflammatory.macrophages has p value = 0.214
Mature.B.cells has p value = 0.582
NK.like.cells has p value = 0.000
Non.inflammatory.macrophages has p value = 0.410
NotDet has p value = 0.987
Periportal.LSECs has p value = 0.464
Portal.endothelial.cells has p value = 0.067
Stellate.cells has p value = 0.006
gamma.delta.T.cells.1 has p value = 0.000
tumor_1 has p value = 0.000
tumor_2 has p value = 0.004
Calculate Cell Frequency and Cell Co-Occurrence in hot/coldspots
[22]:
spot_cellfreq_df, spot_co_occurrence_df = eco.spot_cellfreq(spatial_data=liver_df,
scale=32.0,
library_key='sample',
library_id=library_ids,
spatial_key=['x_slide_um', 'y_slide_um'],
cluster_key='cellType',
spots='hot',
top=None,
selected_comb=None,
restricted=True,
metric='Shannon Diversity')
Processing region: N1 at scale 32.0
1.465 per cent patches are empty
(1009,)
Local spatial stats restricted to tissue region with w of shape(1009, 1009)
Using MoranI
Region N1 contains 6 diversity hotspots
Processing region: N7 at scale 32.0
1.953 per cent patches are empty
(1004,)
Local spatial stats restricted to tissue region with w of shape(1004, 1004)
Using MoranI
Region N7 contains 26 diversity hotspots
Processing region: N9 at scale 32.0
0.000 per cent patches are empty
(1024,)
Local spatial stats restricted to tissue region with w of shape(1024, 1024)
Using MoranI
Region N9 contains 13 diversity hotspots
Processing region: N6 at scale 32.0
0.195 per cent patches are empty
(1022,)
Local spatial stats restricted to tissue region with w of shape(1022, 1022)
Using MoranI
Region N6 contains 13 diversity hotspots
Processing region: N8 at scale 32.0
0.195 per cent patches are empty
(1022,)
Local spatial stats restricted to tissue region with w of shape(1022, 1022)
Using MoranI
Region N8 contains 14 diversity hotspots
Processing region: N2 at scale 32.0
0.000 per cent patches are empty
(1024,)
Local spatial stats restricted to tissue region with w of shape(1024, 1024)
Using MoranI
Region N2 contains 2 diversity hotspots
Processing region: N4 at scale 32.0
0.000 per cent patches are empty
(1024,)
Local spatial stats restricted to tissue region with w of shape(1024, 1024)
Using MoranI
Region N4 contains 24 diversity hotspots
Processing region: N5 at scale 32.0
0.098 per cent patches are empty
(1023,)
Local spatial stats restricted to tissue region with w of shape(1023, 1023)
Using MoranI
Region N5 contains 26 diversity hotspots
Processing region: N0 at scale 32.0
0.684 per cent patches are empty
(1017,)
Local spatial stats restricted to tissue region with w of shape(1017, 1017)
Using MoranI
Region N0 contains 30 diversity hotspots
Processing region: N3 at scale 32.0
0.195 per cent patches are empty
(1022,)
Local spatial stats restricted to tissue region with w of shape(1022, 1022)
Using MoranI
Region N3 contains 3 diversity hotspots
Processing region: C4 at scale 32.0
0.000 per cent patches are empty
(1024,)
Local spatial stats restricted to tissue region with w of shape(1024, 1024)
Using MoranI
Region C4 contains 77 diversity hotspots
Processing region: C0 at scale 32.0
0.000 per cent patches are empty
(1024,)
Local spatial stats restricted to tissue region with w of shape(1024, 1024)
Using MoranI
Region C0 contains 15 diversity hotspots
Processing region: C1 at scale 32.0
1.953 per cent patches are empty
(1004,)
Local spatial stats restricted to tissue region with w of shape(1004, 1004)
Using MoranI
Region C1 contains 17 diversity hotspots
Processing region: C5 at scale 32.0
0.000 per cent patches are empty
(1024,)
Local spatial stats restricted to tissue region with w of shape(1024, 1024)
Using MoranI
Region C5 contains 47 diversity hotspots
Processing region: C6 at scale 32.0
4.297 per cent patches are empty
(980,)
Local spatial stats restricted to tissue region with w of shape(980, 980)
Using MoranI
Region C6 contains 67 diversity hotspots
Processing region: C7 at scale 32.0
0.000 per cent patches are empty
(1024,)
Local spatial stats restricted to tissue region with w of shape(1024, 1024)
Using MoranI
Region C7 contains 98 diversity hotspots
Processing region: C9 at scale 32.0
1.953 per cent patches are empty
(1004,)
Local spatial stats restricted to tissue region with w of shape(1004, 1004)
Using MoranI
Region C9 contains 19 diversity hotspots
Processing region: C2 at scale 32.0
0.000 per cent patches are empty
(1024,)
Local spatial stats restricted to tissue region with w of shape(1024, 1024)
Using MoranI
Region C2 contains 26 diversity hotspots
Processing region: C3 at scale 32.0
2.930 per cent patches are empty
(994,)
Local spatial stats restricted to tissue region with w of shape(994, 994)
Using MoranI
Region C3 contains 81 diversity hotspots
Processing region: C8 at scale 32.0
/Users/Emrys/MESA/docs/notebooks/tutorials/../../../mesa/ecospatial/_ecospatial.py:111: UserWarning: The weights matrix is not fully connected:
There are 2 disconnected components.
There is 1 island with id: 46.
w_new = weights.W(neighbors=new_neighbors, weights=new_weights)
6.348 per cent patches are empty
(959,)
Local spatial stats restricted to tissue region with w of shape(959, 959)
Using MoranI
Region C8 contains 25 diversity hotspots
/Users/Emrys/MESA/docs/notebooks/tutorials/../../../mesa/ecospatial/_ecospatial.py:111: UserWarning: The weights matrix is not fully connected:
There are 3 disconnected components.
There are 2 islands with ids: 31, 35.
w_new = weights.W(neighbors=new_neighbors, weights=new_weights)
[24]:
spot_cellfreq_df['Condition'] = ' '
if 'NotDet' in spot_cellfreq_df.columns:
spot_cellfreq_df = spot_cellfreq_df.drop(columns='NotDet')
spot_cellfreq_df.loc[spot_cellfreq_df.index.str.contains('N'), 'Condition'] = 'Healthy'
spot_cellfreq_df.loc[spot_cellfreq_df.index.str.contains('C'), 'Condition'] = 'HCC'
spot_co_occurrence_subcols = spot_co_occurrence_df.loc[:,spot_co_occurrence_df.mean()>0.05].columns.tolist()
spot_co_occurrence_df['Condition'] = ' '
spot_co_occurrence_df['sample'] = spot_co_occurrence_df.index
spot_co_occurrence_df.loc[spot_co_occurrence_df.index.str.contains('N'), 'Condition'] = 'Healthy'
spot_co_occurrence_df.loc[spot_co_occurrence_df.index.str.contains('C'), 'Condition'] = 'HCC'
spot_co_occurrence_subcols.extend([('Condition',''),('sample','')])
[25]:
spot_cellfreq_df['sample'] = spot_cellfreq_df.index
# Melt the DataFrame
spot_cellfreq_df_melt = spot_cellfreq_df.melt(id_vars=['sample', 'Condition'], var_name='CellType', value_name='Frequency')
[26]:
spot_cellfreq_df_melt
[26]:
| sample | Condition | CellType | Frequency | |
|---|---|---|---|---|
| 0 | N1 | Healthy | Antibody.secreting.B.cells | 0.000000 |
| 1 | N7 | Healthy | Antibody.secreting.B.cells | 0.042184 |
| 2 | N9 | Healthy | Antibody.secreting.B.cells | 0.028409 |
| 3 | N6 | Healthy | Antibody.secreting.B.cells | 0.040201 |
| 4 | N8 | Healthy | Antibody.secreting.B.cells | 0.012658 |
| ... | ... | ... | ... | ... |
| 415 | C7 | HCC | tumor_2 | 0.043659 |
| 416 | C9 | HCC | tumor_2 | 0.091346 |
| 417 | C2 | HCC | tumor_2 | 0.207407 |
| 418 | C3 | HCC | tumor_2 | 0.125506 |
| 419 | C8 | HCC | tumor_2 | 0.000000 |
420 rows × 4 columns
[27]:
selected_cell_types = sorted(spot_cellfreq_df_melt['CellType'].unique())
selected_p_values = []
# Perform t-tests
print(f"p-value before correction:")
for ct in selected_cell_types: # df_melted['CellType'].unique():
subset = spot_cellfreq_df_melt[spot_cellfreq_df_melt['CellType'] == ct]
group1 = subset[subset['Condition'] == 'Healthy']['Frequency']
group2 = subset[subset['Condition'] == 'HCC']['Frequency']
t_stat, p_value = stats.ttest_ind(group1, group2, equal_var=False)
print(f"{ct} has p value = {p_value:.4f}")
selected_p_values.append(p_value)
# Filter the dataframe based on selected CellTypes
df_filtered = spot_cellfreq_df_melt[spot_cellfreq_df_melt['CellType'].isin(selected_cell_types)]
# Plot the filtered data
fig, ax = plt.subplots(figsize=(30,10))
sns.boxplot(data=df_filtered, x='CellType', y='Frequency', hue='Condition', palette='muted', boxprops=dict(alpha=.3), ax=ax, dodge=True,order=selected_cell_types)
sns.swarmplot(data=df_filtered, x='CellType', y='Frequency', hue='Condition', palette='dark:black', size=3.0, dodge=True, order=selected_cell_types, ax=ax, edgecolor='auto', linewidth=0.5)
handles, labels = ax.get_legend_handles_labels()
ax.legend(handles[:2], labels[:2], title="Groups", handletextpad=1, columnspacing=1, bbox_to_anchor=(1, 1), ncol=3, frameon=True)
spot_pvals_corrected = stats.false_discovery_control(selected_p_values, method='bh')
spot_pvals_corrected = dict(map(lambda i,j : (i,j) , selected_cell_types, spot_pvals_corrected))
print('-'*42)
print(f"p-values after correction: ")
yrange = ax.get_ylim()[1] - ax.get_ylim()[0]
for i, ct in enumerate(selected_cell_types):
ax.text(i, yrange, f"p = {spot_pvals_corrected[ct]:.3f}", ha='center', fontsize=12, rotation=90)
print(f"{ct} in hot spots has p value = {spot_pvals_corrected[ct]:.3f}", flush=True)
if spot_pvals_corrected[ct] < 0.05 and p_vals_corrected_dict[ct] > 0.05:
print(f"{ct} in whole tissue has p value = {p_vals_corrected_dict[ct]:.3f}", flush=True)
print('*'*42)
for i in range(len(selected_cell_types) - 1):
ax.axvline(i + 0.55, color='grey', linestyle='--', linewidth=0.5)
ax.set_ylabel("Frequency", fontsize=14)
ax.set_xlabel('')
plt.xticks(rotation=90)
plt.yticks(rotation=90)
plt.show()
plt.close(fig)
p-value before correction:
Antibody.secreting.B.cells has p value = 0.6141
CD3+.alpha.beta.T.cells has p value = 0.0936
Central.venous.LSECs has p value = 0.0433
Cholangiocytes has p value = 0.0485
Erthyroid.cells has p value = 0.0451
Hep has p value = 0.2253
Hep.1 has p value = 0.0046
Hep.3 has p value = 0.0001
Hep.4 has p value = 0.0001
Hep.5 has p value = 0.0002
Hep.6 has p value = 0.0055
Inflammatory.macrophages has p value = 0.0018
Mature.B.cells has p value = 0.0876
NK.like.cells has p value = 0.0342
Non.inflammatory.macrophages has p value = 0.2004
Periportal.LSECs has p value = 0.5968
Portal.endothelial.cells has p value = 0.5409
Stellate.cells has p value = 0.3656
gamma.delta.T.cells.1 has p value = 0.4125
tumor_1 has p value = 0.0000
tumor_2 has p value = 0.0020
------------------------------------------
p-values after correction:
Antibody.secreting.B.cells in hot spots has p value = 0.614
CD3+.alpha.beta.T.cells in hot spots has p value = 0.140
Central.venous.LSECs in hot spots has p value = 0.085
Cholangiocytes in hot spots has p value = 0.085
Erthyroid.cells in hot spots has p value = 0.085
Hep in hot spots has p value = 0.296
Hep.1 in hot spots has p value = 0.014
Hep.3 in hot spots has p value = 0.001
Hep.4 in hot spots has p value = 0.001
Hep.5 in hot spots has p value = 0.001
Hep.6 in hot spots has p value = 0.014
Inflammatory.macrophages in hot spots has p value = 0.007
Inflammatory.macrophages in whole tissue has p value = 0.214
******************************************
Mature.B.cells in hot spots has p value = 0.140
NK.like.cells in hot spots has p value = 0.080
Non.inflammatory.macrophages in hot spots has p value = 0.281
Periportal.LSECs in hot spots has p value = 0.614
Portal.endothelial.cells in hot spots has p value = 0.598
Stellate.cells in hot spots has p value = 0.452
gamma.delta.T.cells.1 in hot spots has p value = 0.481
tumor_1 in hot spots has p value = 0.000
tumor_2 in hot spots has p value = 0.007
Process Cell Co-Occurrence Dataframe
[28]:
union_cols = set(global_co_occurrence_subcols).union(set(spot_co_occurrence_subcols))
[29]:
# Make them have the same set of columns
global_co_occurrence_df = global_co_occurrence_df.reindex(columns=union_cols).fillna(0)
spot_co_occurrence_df = spot_co_occurrence_df.reindex(columns=union_cols).fillna(0)
[30]:
# Global Cell Co-Occurrence
# Multi-index to single-index column
new_columns = []
for col in global_co_occurrence_df.columns:
if isinstance(col, tuple): # This checks if the column is a MultiIndex
# Join only if the column name is not 'Mouse' or 'Condition'
if "sample" not in col and "Condition" not in col:
new_columns.append('&'.join(map(str, col)).strip())
else:
# If 'Mouse' or 'Condition' is in the column, it is not joined with '&'
new_columns.append(col[0])
else:
new_columns.append(col)
global_co_occurrence_df_single = global_co_occurrence_df.copy()
global_co_occurrence_df_single.columns = new_columns
global_co_occurrence_df_single = global_co_occurrence_df_single[[col for col in global_co_occurrence_df_single.columns if 'NotDet' not in col]]
# Melt the DataFrame
global_co_occurrence_melted = global_co_occurrence_df_single.melt(id_vars=['sample', 'Condition'], var_name='Cell Combination', value_name='Frequency')
global_co_occurrence_melted
[30]:
| sample | Condition | Cell Combination | Frequency | |
|---|---|---|---|---|
| 0 | N1 | Healthy | Periportal.LSECs&tumor_1 | 0.0 |
| 1 | N7 | Healthy | Periportal.LSECs&tumor_1 | 0.0 |
| 2 | N9 | Healthy | Periportal.LSECs&tumor_1 | 0.0 |
| 3 | N6 | Healthy | Periportal.LSECs&tumor_1 | 0.0 |
| 4 | N8 | Healthy | Periportal.LSECs&tumor_1 | 0.0 |
| ... | ... | ... | ... | ... |
| 1555 | C7 | HCC | CD3+.alpha.beta.T.cells&Hep.6 | 0.0 |
| 1556 | C9 | HCC | CD3+.alpha.beta.T.cells&Hep.6 | 0.0 |
| 1557 | C2 | HCC | CD3+.alpha.beta.T.cells&Hep.6 | 0.0 |
| 1558 | C3 | HCC | CD3+.alpha.beta.T.cells&Hep.6 | 0.0 |
| 1559 | C8 | HCC | CD3+.alpha.beta.T.cells&Hep.6 | 0.0 |
1560 rows × 4 columns
[31]:
# Global Cell Co-Occurrence
selected_cell_types = sorted(global_co_occurrence_melted['Cell Combination'].unique())
selected_p_values = []
# Perform t-tests
print(f"p-value before correction:")
for ct in selected_cell_types:
subset = global_co_occurrence_melted[global_co_occurrence_melted['Cell Combination'] == ct]
group1 = subset[subset['Condition'] == 'Healthy']['Frequency']
group2 = subset[subset['Condition'] == 'HCC']['Frequency']
t_stat, p_value = stats.ttest_ind(group1, group2, equal_var=False)
print(f"{ct} has p value = {p_value:.3f}")
selected_p_values.append(p_value)
# Filter the dataframe based on selected Cell Combinations
df_filtered = global_co_occurrence_melted[global_co_occurrence_melted['Cell Combination'].isin(selected_cell_types)]
# Plot the filtered data
fig, ax = plt.subplots(figsize=(45,10))
sns.boxplot(data=df_filtered, x='Cell Combination', y='Frequency', hue='Condition', palette='muted', boxprops=dict(alpha=.3), ax=ax, dodge=True,order=selected_cell_types)
sns.swarmplot(data=df_filtered, x='Cell Combination', y='Frequency', hue='Condition', palette='dark:black', size=1.0, dodge=True,order=selected_cell_types, ax=ax, edgecolor='gray', linewidth=0.5)
handles, labels = ax.get_legend_handles_labels()
ax.legend(handles[:2], labels[:2], title="Groups", handletextpad=1, columnspacing=1, bbox_to_anchor=(1, 1), ncol=3, frameon=True)
pvals_corrected = stats.false_discovery_control(selected_p_values, method='bh')
print('-'*42)
print(f"p-values after correction:")
p_vals_corrected_dict = {}
yrange = ax.get_ylim()[1] - ax.get_ylim()[0]
for i, ct in enumerate(selected_cell_types):
ax.text(i, yrange, f"p = {pvals_corrected[i]:.3f}", ha='center', fontsize=8, rotation=0)
print(f"{ct} has p value = {pvals_corrected[i]:.3f}", flush=True)
p_vals_corrected_dict[ct] = pvals_corrected[i]
for i in range(len(selected_cell_types) - 1):
ax.axvline(i + 0.55, color='grey', linestyle='--', linewidth=0.5)
ax.set_ylabel("Frequency", fontsize=14)
ax.set_xlabel('')
plt.xticks(rotation=90)
plt.show()
p-value before correction:
Antibody.secreting.B.cells&CD3+.alpha.beta.T.cells has p value = 0.962
Antibody.secreting.B.cells&Inflammatory.macrophages has p value = 0.734
Antibody.secreting.B.cells&Mature.B.cells has p value = 0.560
Antibody.secreting.B.cells&Periportal.LSECs has p value = 0.111
Antibody.secreting.B.cells&Stellate.cells has p value = 0.406
CD3+.alpha.beta.T.cells&Central.venous.LSECs has p value = 0.001
CD3+.alpha.beta.T.cells&Cholangiocytes has p value = 0.000
CD3+.alpha.beta.T.cells&Hep.1 has p value = 0.000
CD3+.alpha.beta.T.cells&Hep.3 has p value = 0.000
CD3+.alpha.beta.T.cells&Hep.4 has p value = 0.000
CD3+.alpha.beta.T.cells&Hep.5 has p value = 0.000
CD3+.alpha.beta.T.cells&Hep.6 has p value = 0.000
CD3+.alpha.beta.T.cells&Inflammatory.macrophages has p value = 0.204
CD3+.alpha.beta.T.cells&Mature.B.cells has p value = 0.420
CD3+.alpha.beta.T.cells&NK.like.cells has p value = 0.004
CD3+.alpha.beta.T.cells&Non.inflammatory.macrophages has p value = 0.945
CD3+.alpha.beta.T.cells&Periportal.LSECs has p value = 0.044
CD3+.alpha.beta.T.cells&Stellate.cells has p value = 0.123
CD3+.alpha.beta.T.cells&tumor_1 has p value = 0.000
CD3+.alpha.beta.T.cells&tumor_2 has p value = 0.000
Central.venous.LSECs&Hep.1 has p value = 0.000
Central.venous.LSECs&Hep.3 has p value = 0.000
Central.venous.LSECs&Hep.4 has p value = 0.000
Central.venous.LSECs&Hep.5 has p value = 0.000
Central.venous.LSECs&Inflammatory.macrophages has p value = 0.223
Central.venous.LSECs&Non.inflammatory.macrophages has p value = 0.000
Central.venous.LSECs&Stellate.cells has p value = 0.000
Cholangiocytes&Hep.4 has p value = 0.000
Cholangiocytes&Inflammatory.macrophages has p value = 0.001
Cholangiocytes&Mature.B.cells has p value = 0.001
Cholangiocytes&Periportal.LSECs has p value = 0.000
Cholangiocytes&Stellate.cells has p value = 0.000
Hep.1&Hep.3 has p value = 0.000
Hep.1&Hep.4 has p value = 0.000
Hep.1&Hep.5 has p value = 0.000
Hep.1&Hep.6 has p value = 0.000
Hep.1&Inflammatory.macrophages has p value = 0.000
Hep.1&Non.inflammatory.macrophages has p value = 0.000
Hep.1&Stellate.cells has p value = 0.000
Hep.3&Hep.4 has p value = 0.000
Hep.3&Hep.5 has p value = 0.000
Hep.3&Hep.6 has p value = 0.000
Hep.3&Inflammatory.macrophages has p value = 0.000
Hep.3&Mature.B.cells has p value = 0.000
Hep.3&Non.inflammatory.macrophages has p value = 0.000
Hep.3&Periportal.LSECs has p value = 0.000
Hep.3&Stellate.cells has p value = 0.000
Hep.4&Hep.5 has p value = 0.000
Hep.4&Hep.6 has p value = 0.000
Hep.4&Inflammatory.macrophages has p value = 0.000
Hep.4&Mature.B.cells has p value = 0.000
Hep.4&Non.inflammatory.macrophages has p value = 0.000
Hep.4&Periportal.LSECs has p value = 0.000
Hep.4&Stellate.cells has p value = 0.000
Hep.5&Hep.6 has p value = 0.000
Hep.5&Inflammatory.macrophages has p value = 0.000
Hep.5&Non.inflammatory.macrophages has p value = 0.000
Hep.5&Stellate.cells has p value = 0.000
Inflammatory.macrophages&Mature.B.cells has p value = 0.157
Inflammatory.macrophages&Non.inflammatory.macrophages has p value = 0.213
Inflammatory.macrophages&Periportal.LSECs has p value = 0.013
Inflammatory.macrophages&Stellate.cells has p value = 0.790
Inflammatory.macrophages&tumor_1 has p value = 0.000
Inflammatory.macrophages&tumor_2 has p value = 0.000
Mature.B.cells&Non.inflammatory.macrophages has p value = 0.337
Mature.B.cells&Periportal.LSECs has p value = 0.072
Mature.B.cells&Stellate.cells has p value = 0.863
Mature.B.cells&tumor_1 has p value = 0.000
NK.like.cells&Stellate.cells has p value = 0.000
Non.inflammatory.macrophages&Periportal.LSECs has p value = 0.004
Non.inflammatory.macrophages&Stellate.cells has p value = 0.201
Non.inflammatory.macrophages&tumor_1 has p value = 0.000
Periportal.LSECs&Stellate.cells has p value = 0.616
Periportal.LSECs&tumor_1 has p value = 0.000
Periportal.LSECs&tumor_2 has p value = 0.001
Stellate.cells&tumor_1 has p value = 0.000
Stellate.cells&tumor_2 has p value = 0.000
tumor_1&tumor_2 has p value = 0.001
/var/folders/7g/phdhh_ld3dlbnrst0t60bwzr0000gn/T/ipykernel_73139/917667374.py:22: FutureWarning: Use "auto" to set automatic grayscale colors. From v0.14.0, "gray" will default to matplotlib's definition.
sns.swarmplot(data=df_filtered, x='Cell Combination', y='Frequency', hue='Condition', palette='dark:black', size=1.0, dodge=True,order=selected_cell_types, ax=ax, edgecolor='gray', linewidth=0.5)
------------------------------------------
p-values after correction:
Antibody.secreting.B.cells&CD3+.alpha.beta.T.cells has p value = 0.962
Antibody.secreting.B.cells&Inflammatory.macrophages has p value = 0.773
Antibody.secreting.B.cells&Mature.B.cells has p value = 0.607
Antibody.secreting.B.cells&Periportal.LSECs has p value = 0.139
Antibody.secreting.B.cells&Stellate.cells has p value = 0.452
CD3+.alpha.beta.T.cells&Central.venous.LSECs has p value = 0.001
CD3+.alpha.beta.T.cells&Cholangiocytes has p value = 0.000
CD3+.alpha.beta.T.cells&Hep.1 has p value = 0.000
CD3+.alpha.beta.T.cells&Hep.3 has p value = 0.000
CD3+.alpha.beta.T.cells&Hep.4 has p value = 0.000
CD3+.alpha.beta.T.cells&Hep.5 has p value = 0.000
CD3+.alpha.beta.T.cells&Hep.6 has p value = 0.000
CD3+.alpha.beta.T.cells&Inflammatory.macrophages has p value = 0.241
CD3+.alpha.beta.T.cells&Mature.B.cells has p value = 0.461
CD3+.alpha.beta.T.cells&NK.like.cells has p value = 0.005
CD3+.alpha.beta.T.cells&Non.inflammatory.macrophages has p value = 0.957
CD3+.alpha.beta.T.cells&Periportal.LSECs has p value = 0.057
CD3+.alpha.beta.T.cells&Stellate.cells has p value = 0.152
CD3+.alpha.beta.T.cells&tumor_1 has p value = 0.000
CD3+.alpha.beta.T.cells&tumor_2 has p value = 0.000
Central.venous.LSECs&Hep.1 has p value = 0.000
Central.venous.LSECs&Hep.3 has p value = 0.000
Central.venous.LSECs&Hep.4 has p value = 0.000
Central.venous.LSECs&Hep.5 has p value = 0.000
Central.venous.LSECs&Inflammatory.macrophages has p value = 0.255
Central.venous.LSECs&Non.inflammatory.macrophages has p value = 0.000
Central.venous.LSECs&Stellate.cells has p value = 0.000
Cholangiocytes&Hep.4 has p value = 0.000
Cholangiocytes&Inflammatory.macrophages has p value = 0.001
Cholangiocytes&Mature.B.cells has p value = 0.002
Cholangiocytes&Periportal.LSECs has p value = 0.000
Cholangiocytes&Stellate.cells has p value = 0.000
Hep.1&Hep.3 has p value = 0.000
Hep.1&Hep.4 has p value = 0.000
Hep.1&Hep.5 has p value = 0.000
Hep.1&Hep.6 has p value = 0.000
Hep.1&Inflammatory.macrophages has p value = 0.000
Hep.1&Non.inflammatory.macrophages has p value = 0.000
Hep.1&Stellate.cells has p value = 0.000
Hep.3&Hep.4 has p value = 0.000
Hep.3&Hep.5 has p value = 0.000
Hep.3&Hep.6 has p value = 0.000
Hep.3&Inflammatory.macrophages has p value = 0.000
Hep.3&Mature.B.cells has p value = 0.000
Hep.3&Non.inflammatory.macrophages has p value = 0.000
Hep.3&Periportal.LSECs has p value = 0.000
Hep.3&Stellate.cells has p value = 0.000
Hep.4&Hep.5 has p value = 0.000
Hep.4&Hep.6 has p value = 0.000
Hep.4&Inflammatory.macrophages has p value = 0.000
Hep.4&Mature.B.cells has p value = 0.000
Hep.4&Non.inflammatory.macrophages has p value = 0.000
Hep.4&Periportal.LSECs has p value = 0.000
Hep.4&Stellate.cells has p value = 0.000
Hep.5&Hep.6 has p value = 0.000
Hep.5&Inflammatory.macrophages has p value = 0.000
Hep.5&Non.inflammatory.macrophages has p value = 0.000
Hep.5&Stellate.cells has p value = 0.000
Inflammatory.macrophages&Mature.B.cells has p value = 0.192
Inflammatory.macrophages&Non.inflammatory.macrophages has p value = 0.248
Inflammatory.macrophages&Periportal.LSECs has p value = 0.017
Inflammatory.macrophages&Stellate.cells has p value = 0.821
Inflammatory.macrophages&tumor_1 has p value = 0.000
Inflammatory.macrophages&tumor_2 has p value = 0.000
Mature.B.cells&Non.inflammatory.macrophages has p value = 0.381
Mature.B.cells&Periportal.LSECs has p value = 0.092
Mature.B.cells&Stellate.cells has p value = 0.885
Mature.B.cells&tumor_1 has p value = 0.000
NK.like.cells&Stellate.cells has p value = 0.000
Non.inflammatory.macrophages&Periportal.LSECs has p value = 0.005
Non.inflammatory.macrophages&Stellate.cells has p value = 0.241
Non.inflammatory.macrophages&tumor_1 has p value = 0.000
Periportal.LSECs&Stellate.cells has p value = 0.658
Periportal.LSECs&tumor_1 has p value = 0.000
Periportal.LSECs&tumor_2 has p value = 0.001
Stellate.cells&tumor_1 has p value = 0.000
Stellate.cells&tumor_2 has p value = 0.001
tumor_1&tumor_2 has p value = 0.001
[32]:
# Spot Cell Co-Occurrence
# Multi-index to single-index column
new_columns = []
for col in spot_co_occurrence_df.columns:
if isinstance(col, tuple): # This checks if the column is a MultiIndex
# Join only if the column name is not 'Mouse' or 'Condition'
if "sample" not in col and "Condition" not in col:
new_columns.append('&'.join(map(str, col)).strip())
else:
# If 'Mouse' or 'Condition' is in the column, it is not joined with '&'
new_columns.append(col[0])
else:
new_columns.append(col)
spot_co_occurrence_df_single = spot_co_occurrence_df.copy()
spot_co_occurrence_df_single.columns = new_columns
spot_co_occurrence_df_single = spot_co_occurrence_df_single[[col for col in spot_co_occurrence_df_single.columns if 'noid' not in col]]
# Melt the DataFrame
spot_co_occurrence_melted = spot_co_occurrence_df_single.melt(id_vars=['sample', 'Condition'], var_name='Cell Combination', value_name='Frequency')
spot_co_occurrence_melted
[32]:
| sample | Condition | Cell Combination | Frequency | |
|---|---|---|---|---|
| 0 | N1 | Healthy | Periportal.LSECs&tumor_1 | 0.0 |
| 1 | N7 | Healthy | Periportal.LSECs&tumor_1 | 0.0 |
| 2 | N9 | Healthy | Periportal.LSECs&tumor_1 | 0.0 |
| 3 | N6 | Healthy | Periportal.LSECs&tumor_1 | 0.0 |
| 4 | N8 | Healthy | Periportal.LSECs&tumor_1 | 0.0 |
| ... | ... | ... | ... | ... |
| 1555 | C7 | HCC | CD3+.alpha.beta.T.cells&Hep.6 | 0.0 |
| 1556 | C9 | HCC | CD3+.alpha.beta.T.cells&Hep.6 | 0.0 |
| 1557 | C2 | HCC | CD3+.alpha.beta.T.cells&Hep.6 | 0.0 |
| 1558 | C3 | HCC | CD3+.alpha.beta.T.cells&Hep.6 | 0.0 |
| 1559 | C8 | HCC | CD3+.alpha.beta.T.cells&Hep.6 | 0.0 |
1560 rows × 4 columns
[33]:
# Spot Cell Co-Occurrence
selected_cell_types = sorted(spot_co_occurrence_melted['Cell Combination'].unique())
selected_p_values = []
# Perform t-tests
print(f"p-value before correction: ")
for ct in selected_cell_types: # df_melted['CellType'].unique():
subset = spot_co_occurrence_melted[spot_co_occurrence_melted['Cell Combination'] == ct]
group1 = subset[subset['Condition'] == 'Healthy']['Frequency']
group2 = subset[subset['Condition'] == 'HCC']['Frequency']
t_stat, p_value = stats.ttest_ind(group1, group2, equal_var=False)
print(f"{ct} has p value = {p_value:.3f}")
selected_p_values.append(p_value)
# Filter the dataframe based on selected CellTypes
df_filtered = spot_co_occurrence_melted[spot_co_occurrence_melted['Cell Combination'].isin(selected_cell_types)]
# Plot the filtered data
fig, ax = plt.subplots(figsize=(42,10))
sns.boxplot(data=df_filtered, x='Cell Combination', y='Frequency', hue='Condition', palette='muted', boxprops=dict(alpha=.3), ax=ax, dodge=True,order=selected_cell_types)
sns.swarmplot(data=df_filtered, x='Cell Combination', y='Frequency', hue='Condition', palette='dark:black', size=2.0, dodge=True,order=selected_cell_types, ax=ax, edgecolor='gray', linewidth=0.5)
handles, labels = ax.get_legend_handles_labels()
ax.legend(handles[:2], labels[:2], title="Groups", handletextpad=1, columnspacing=1, bbox_to_anchor=(1, 1), ncol=3, frameon=True)
hot_pvals_corrected = stats.false_discovery_control(selected_p_values, method='bh')
print('-'*42)
print(f"p-values after correction:")
highlighted_comb = []
yrange = ax.get_ylim()[1] - ax.get_ylim()[0]
for i, ct in enumerate(selected_cell_types):
ax.text(i, yrange, f"p = {pvals_corrected[i]:.3f}", ha='center', fontsize=8, rotation=0)
print(f"{ct} in hot spots has p value = {hot_pvals_corrected[i]:.3f}", flush=True)
if hot_pvals_corrected[i] < 0.05 and p_vals_corrected_dict[ct] >= 0.05:
highlighted_comb.append(tuple(map(str.strip, ct.split('&'))))
print(f"{ct} in whole tissue has p value = {p_vals_corrected_dict[ct]:.3f}", flush=True)
print('*'*42)
for i in range(len(selected_cell_types) - 1):
ax.axvline(i + 0.55, color='grey', linestyle='--', linewidth=0.5)
ax.set_ylabel("Frequency", fontsize=14)
ax.set_xlabel('')
plt.xticks(rotation=90)
plt.yticks(rotation=90)
plt.show()
p-value before correction:
Antibody.secreting.B.cells&CD3+.alpha.beta.T.cells has p value = 0.293
Antibody.secreting.B.cells&Inflammatory.macrophages has p value = 0.400
Antibody.secreting.B.cells&Mature.B.cells has p value = 0.456
Antibody.secreting.B.cells&Periportal.LSECs has p value = 0.092
Antibody.secreting.B.cells&Stellate.cells has p value = 0.211
CD3+.alpha.beta.T.cells&Central.venous.LSECs has p value = 0.069
CD3+.alpha.beta.T.cells&Cholangiocytes has p value = 0.037
CD3+.alpha.beta.T.cells&Hep.1 has p value = 0.001
CD3+.alpha.beta.T.cells&Hep.3 has p value = 0.000
CD3+.alpha.beta.T.cells&Hep.4 has p value = 0.000
CD3+.alpha.beta.T.cells&Hep.5 has p value = 0.000
CD3+.alpha.beta.T.cells&Hep.6 has p value = 0.023
CD3+.alpha.beta.T.cells&Inflammatory.macrophages has p value = 0.060
CD3+.alpha.beta.T.cells&Mature.B.cells has p value = 0.266
CD3+.alpha.beta.T.cells&NK.like.cells has p value = 0.016
CD3+.alpha.beta.T.cells&Non.inflammatory.macrophages has p value = 0.008
CD3+.alpha.beta.T.cells&Periportal.LSECs has p value = 0.928
CD3+.alpha.beta.T.cells&Stellate.cells has p value = 0.569
CD3+.alpha.beta.T.cells&tumor_1 has p value = 0.000
CD3+.alpha.beta.T.cells&tumor_2 has p value = 0.001
Central.venous.LSECs&Hep.1 has p value = 0.034
Central.venous.LSECs&Hep.3 has p value = 0.021
Central.venous.LSECs&Hep.4 has p value = 0.013
Central.venous.LSECs&Hep.5 has p value = 0.004
Central.venous.LSECs&Inflammatory.macrophages has p value = 0.526
Central.venous.LSECs&Non.inflammatory.macrophages has p value = 0.278
Central.venous.LSECs&Stellate.cells has p value = 0.177
Cholangiocytes&Hep.4 has p value = 0.021
Cholangiocytes&Inflammatory.macrophages has p value = 0.104
Cholangiocytes&Mature.B.cells has p value = 0.161
Cholangiocytes&Periportal.LSECs has p value = 0.033
Cholangiocytes&Stellate.cells has p value = 0.035
Hep.1&Hep.3 has p value = 0.002
Hep.1&Hep.4 has p value = 0.001
Hep.1&Hep.5 has p value = 0.000
Hep.1&Hep.6 has p value = 0.009
Hep.1&Inflammatory.macrophages has p value = 0.087
Hep.1&Non.inflammatory.macrophages has p value = 0.053
Hep.1&Stellate.cells has p value = 0.002
Hep.3&Hep.4 has p value = 0.000
Hep.3&Hep.5 has p value = 0.000
Hep.3&Hep.6 has p value = 0.006
Hep.3&Inflammatory.macrophages has p value = 0.005
Hep.3&Mature.B.cells has p value = 0.004
Hep.3&Non.inflammatory.macrophages has p value = 0.027
Hep.3&Periportal.LSECs has p value = 0.009
Hep.3&Stellate.cells has p value = 0.000
Hep.4&Hep.5 has p value = 0.000
Hep.4&Hep.6 has p value = 0.005
Hep.4&Inflammatory.macrophages has p value = 0.005
Hep.4&Mature.B.cells has p value = 0.005
Hep.4&Non.inflammatory.macrophages has p value = 0.019
Hep.4&Periportal.LSECs has p value = 0.025
Hep.4&Stellate.cells has p value = 0.000
Hep.5&Hep.6 has p value = 0.007
Hep.5&Inflammatory.macrophages has p value = 0.002
Hep.5&Non.inflammatory.macrophages has p value = 0.012
Hep.5&Stellate.cells has p value = 0.000
Inflammatory.macrophages&Mature.B.cells has p value = 0.024
Inflammatory.macrophages&Non.inflammatory.macrophages has p value = 0.017
Inflammatory.macrophages&Periportal.LSECs has p value = 0.617
Inflammatory.macrophages&Stellate.cells has p value = 0.252
Inflammatory.macrophages&tumor_1 has p value = 0.000
Inflammatory.macrophages&tumor_2 has p value = 0.001
Mature.B.cells&Non.inflammatory.macrophages has p value = 0.346
Mature.B.cells&Periportal.LSECs has p value = 0.671
Mature.B.cells&Stellate.cells has p value = 0.757
Mature.B.cells&tumor_1 has p value = 0.001
NK.like.cells&Stellate.cells has p value = 0.016
Non.inflammatory.macrophages&Periportal.LSECs has p value = 0.003
Non.inflammatory.macrophages&Stellate.cells has p value = 0.122
Non.inflammatory.macrophages&tumor_1 has p value = 0.000
Periportal.LSECs&Stellate.cells has p value = 0.655
Periportal.LSECs&tumor_1 has p value = 0.000
Periportal.LSECs&tumor_2 has p value = 0.003
Stellate.cells&tumor_1 has p value = 0.000
Stellate.cells&tumor_2 has p value = 0.002
tumor_1&tumor_2 has p value = 0.001
/var/folders/7g/phdhh_ld3dlbnrst0t60bwzr0000gn/T/ipykernel_73139/754806585.py:22: FutureWarning: Use "auto" to set automatic grayscale colors. From v0.14.0, "gray" will default to matplotlib's definition.
sns.swarmplot(data=df_filtered, x='Cell Combination', y='Frequency', hue='Condition', palette='dark:black', size=2.0, dodge=True,order=selected_cell_types, ax=ax, edgecolor='gray', linewidth=0.5)
/opt/miniconda3/envs/mesa/lib/python3.11/site-packages/seaborn/categorical.py:3370: UserWarning: 10.0% of the points cannot be placed; you may want to decrease the size of the markers or use stripplot.
warnings.warn(msg, UserWarning)
------------------------------------------
p-values after correction:
Antibody.secreting.B.cells&CD3+.alpha.beta.T.cells in hot spots has p value = 0.336
Antibody.secreting.B.cells&Inflammatory.macrophages in hot spots has p value = 0.446
Antibody.secreting.B.cells&Mature.B.cells in hot spots has p value = 0.501
Antibody.secreting.B.cells&Periportal.LSECs in hot spots has p value = 0.122
Antibody.secreting.B.cells&Stellate.cells in hot spots has p value = 0.257
CD3+.alpha.beta.T.cells&Central.venous.LSECs in hot spots has p value = 0.094
CD3+.alpha.beta.T.cells&Cholangiocytes in hot spots has p value = 0.053
CD3+.alpha.beta.T.cells&Hep.1 in hot spots has p value = 0.002
CD3+.alpha.beta.T.cells&Hep.3 in hot spots has p value = 0.000
CD3+.alpha.beta.T.cells&Hep.4 in hot spots has p value = 0.000
CD3+.alpha.beta.T.cells&Hep.5 in hot spots has p value = 0.000
CD3+.alpha.beta.T.cells&Hep.6 in hot spots has p value = 0.039
CD3+.alpha.beta.T.cells&Inflammatory.macrophages in hot spots has p value = 0.084
CD3+.alpha.beta.T.cells&Mature.B.cells in hot spots has p value = 0.314
CD3+.alpha.beta.T.cells&NK.like.cells in hot spots has p value = 0.030
CD3+.alpha.beta.T.cells&Non.inflammatory.macrophages in hot spots has p value = 0.018
CD3+.alpha.beta.T.cells&Non.inflammatory.macrophages in whole tissue has p value = 0.957
******************************************
CD3+.alpha.beta.T.cells&Periportal.LSECs in hot spots has p value = 0.928
CD3+.alpha.beta.T.cells&Stellate.cells in hot spots has p value = 0.608
CD3+.alpha.beta.T.cells&tumor_1 in hot spots has p value = 0.000
CD3+.alpha.beta.T.cells&tumor_2 in hot spots has p value = 0.003
Central.venous.LSECs&Hep.1 in hot spots has p value = 0.050
Central.venous.LSECs&Hep.3 in hot spots has p value = 0.036
Central.venous.LSECs&Hep.4 in hot spots has p value = 0.026
Central.venous.LSECs&Hep.5 in hot spots has p value = 0.011
Central.venous.LSECs&Inflammatory.macrophages in hot spots has p value = 0.570
Central.venous.LSECs&Non.inflammatory.macrophages in hot spots has p value = 0.323
Central.venous.LSECs&Stellate.cells in hot spots has p value = 0.219
Cholangiocytes&Hep.4 in hot spots has p value = 0.036
Cholangiocytes&Inflammatory.macrophages in hot spots has p value = 0.136
Cholangiocytes&Mature.B.cells in hot spots has p value = 0.203
Cholangiocytes&Periportal.LSECs in hot spots has p value = 0.050
Cholangiocytes&Stellate.cells in hot spots has p value = 0.052
Hep.1&Hep.3 in hot spots has p value = 0.007
Hep.1&Hep.4 in hot spots has p value = 0.004
Hep.1&Hep.5 in hot spots has p value = 0.002
Hep.1&Hep.6 in hot spots has p value = 0.019
Hep.1&Inflammatory.macrophages in hot spots has p value = 0.117
Hep.1&Non.inflammatory.macrophages in hot spots has p value = 0.075
Hep.1&Stellate.cells in hot spots has p value = 0.006
Hep.3&Hep.4 in hot spots has p value = 0.000
Hep.3&Hep.5 in hot spots has p value = 0.000
Hep.3&Hep.6 in hot spots has p value = 0.014
Hep.3&Inflammatory.macrophages in hot spots has p value = 0.013
Hep.3&Mature.B.cells in hot spots has p value = 0.012
Hep.3&Non.inflammatory.macrophages in hot spots has p value = 0.043
Hep.3&Periportal.LSECs in hot spots has p value = 0.018
Hep.3&Stellate.cells in hot spots has p value = 0.000
Hep.4&Hep.5 in hot spots has p value = 0.000
Hep.4&Hep.6 in hot spots has p value = 0.013
Hep.4&Inflammatory.macrophages in hot spots has p value = 0.013
Hep.4&Mature.B.cells in hot spots has p value = 0.013
Hep.4&Non.inflammatory.macrophages in hot spots has p value = 0.033
Hep.4&Periportal.LSECs in hot spots has p value = 0.040
Hep.4&Stellate.cells in hot spots has p value = 0.000
Hep.5&Hep.6 in hot spots has p value = 0.015
Hep.5&Inflammatory.macrophages in hot spots has p value = 0.006
Hep.5&Non.inflammatory.macrophages in hot spots has p value = 0.024
Hep.5&Stellate.cells in hot spots has p value = 0.000
Inflammatory.macrophages&Mature.B.cells in hot spots has p value = 0.039
Inflammatory.macrophages&Mature.B.cells in whole tissue has p value = 0.192
******************************************
Inflammatory.macrophages&Non.inflammatory.macrophages in hot spots has p value = 0.031
Inflammatory.macrophages&Non.inflammatory.macrophages in whole tissue has p value = 0.248
******************************************
Inflammatory.macrophages&Periportal.LSECs in hot spots has p value = 0.650
Inflammatory.macrophages&Stellate.cells in hot spots has p value = 0.303
Inflammatory.macrophages&tumor_1 in hot spots has p value = 0.000
Inflammatory.macrophages&tumor_2 in hot spots has p value = 0.002
Mature.B.cells&Non.inflammatory.macrophages in hot spots has p value = 0.391
Mature.B.cells&Periportal.LSECs in hot spots has p value = 0.689
Mature.B.cells&Stellate.cells in hot spots has p value = 0.767
Mature.B.cells&tumor_1 in hot spots has p value = 0.004
NK.like.cells&Stellate.cells in hot spots has p value = 0.030
Non.inflammatory.macrophages&Periportal.LSECs in hot spots has p value = 0.010
Non.inflammatory.macrophages&Stellate.cells in hot spots has p value = 0.156
Non.inflammatory.macrophages&tumor_1 in hot spots has p value = 0.000
Periportal.LSECs&Stellate.cells in hot spots has p value = 0.681
Periportal.LSECs&tumor_1 in hot spots has p value = 0.000
Periportal.LSECs&tumor_2 in hot spots has p value = 0.009
Stellate.cells&tumor_1 in hot spots has p value = 0.000
Stellate.cells&tumor_2 in hot spots has p value = 0.006
tumor_1&tumor_2 in hot spots has p value = 0.003
Cell Co-Occurrence Circle Plot
[34]:
circoplot_df1 = global_co_occurrence_df.sort_index(axis=1, level=[0,1]).drop(columns=['sample'])
circoplot_df1 = circoplot_df1[[col for col in circoplot_df1.columns if 'NotDet' not in col]]
# Group by 'Condition' and calculate the mean of the other columns
circoplot_df1 = circoplot_df1.groupby('Condition').mean().reset_index()
circoplot_df1 = circoplot_df1.set_index('Condition')
circoplot_df1
[34]:
| Antibody.secreting.B.cells | CD3+.alpha.beta.T.cells | ... | NK.like.cells | Non.inflammatory.macrophages | Periportal.LSECs | Stellate.cells | tumor_1 | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CD3+.alpha.beta.T.cells | Inflammatory.macrophages | Mature.B.cells | Periportal.LSECs | Stellate.cells | Central.venous.LSECs | Cholangiocytes | Hep.1 | Hep.3 | Hep.4 | ... | Stellate.cells | Periportal.LSECs | Stellate.cells | tumor_1 | Stellate.cells | tumor_1 | tumor_2 | tumor_1 | tumor_2 | tumor_2 | |
| Condition | |||||||||||||||||||||
| HCC | 0.022196 | 0.018992 | 0.016074 | 0.011441 | 0.016042 | 0.023526 | 0.006582 | 0.000000 | 0.000000 | 0.000000 | ... | 0.006711 | 0.065835 | 0.058693 | 0.153495 | 0.080329 | 0.195998 | 0.034058 | 0.127627 | 0.027149 | 0.167421 |
| Healthy | 0.021499 | 0.014525 | 0.009815 | 0.023953 | 0.025427 | 0.046878 | 0.043016 | 0.091162 | 0.179436 | 0.305974 | ... | 0.034734 | 0.006585 | 0.096116 | 0.000000 | 0.091030 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | 0.000000 |
2 rows × 78 columns
[35]:
circoplot_df2 = global_cellfreq_df.drop(columns=['NotDet'])
# Group by 'Condition' and calculate the mean of the other columns
circoplot_df2 = circoplot_df2.groupby('Condition').mean().reset_index()
circoplot_df2 = circoplot_df2.set_index('Condition')
circoplot_df2
[35]:
| cellType | Antibody.secreting.B.cells | CD3+.alpha.beta.T.cells | Central.venous.LSECs | Cholangiocytes | Erthyroid.cells | Hep | Hep.1 | Hep.3 | Hep.4 | Hep.5 | ... | Inflammatory.macrophages | Mature.B.cells | NK.like.cells | Non.inflammatory.macrophages | Periportal.LSECs | Portal.endothelial.cells | Stellate.cells | gamma.delta.T.cells.1 | tumor_1 | tumor_2 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Condition | |||||||||||||||||||||
| HCC | 0.004335 | 0.052099 | 0.004248 | 0.001246 | 0.000355 | 0.004334 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | ... | 0.041426 | 0.012241 | 0.001372 | 0.021900 | 0.029793 | 0.002345 | 0.024236 | 0.001431 | 0.743387 | 0.055246 |
| Healthy | 0.005318 | 0.042423 | 0.012634 | 0.018712 | 0.003344 | 0.000000 | 0.041616 | 0.068025 | 0.368931 | 0.263174 | ... | 0.019289 | 0.008970 | 0.008195 | 0.028868 | 0.024384 | 0.001588 | 0.055661 | 0.009370 | 0.000000 | 0.000000 |
2 rows × 21 columns
[37]:
patient_group = 'HCC'
eco.create_circos_plot(circoplot_df1.loc[[patient_group]],
cell_type_colors_hex=None,
cell_abundance=circoplot_df2.loc[[patient_group]],
threshold=0.05,
edge_weights_scaler=10,
highlighted_edges=None,
node_weights_scaler=5000,
figure_size=(8,8),
save_path='/Users/Emrys/Downloads/'+patient_group+'_wholetissue_circoplot.svg')
[38]:
circoplot_df1 = spot_co_occurrence_df.sort_index(axis=1, level=[0,1]).drop(columns=['sample'])
circoplot_df1 = circoplot_df1[[col for col in circoplot_df1.columns if 'NotDet' not in col]]
# Group by 'Condition' and calculate the mean of the other columns
circoplot_df1 = circoplot_df1.groupby('Condition').mean().reset_index()
circoplot_df1 = circoplot_df1.set_index('Condition')
circoplot_df1
[38]:
| Antibody.secreting.B.cells | CD3+.alpha.beta.T.cells | ... | NK.like.cells | Non.inflammatory.macrophages | Periportal.LSECs | Stellate.cells | tumor_1 | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CD3+.alpha.beta.T.cells | Inflammatory.macrophages | Mature.B.cells | Periportal.LSECs | Stellate.cells | Central.venous.LSECs | Cholangiocytes | Hep.1 | Hep.3 | Hep.4 | ... | Stellate.cells | Periportal.LSECs | Stellate.cells | tumor_1 | Stellate.cells | tumor_1 | tumor_2 | tumor_1 | tumor_2 | tumor_2 | |
| Condition | |||||||||||||||||||||
| HCC | 0.071472 | 0.054681 | 0.048080 | 0.041909 | 0.060433 | 0.061259 | 0.042978 | 0.000000 | 0.000000 | 0.000000 | ... | 0.015938 | 0.106321 | 0.180634 | 0.213524 | 0.280178 | 0.363789 | 0.146445 | 0.40268 | 0.134935 | 0.35633 |
| Healthy | 0.133004 | 0.090733 | 0.075861 | 0.144542 | 0.136850 | 0.171346 | 0.195916 | 0.267537 | 0.408425 | 0.444679 | ... | 0.104890 | 0.007692 | 0.072244 | 0.000000 | 0.324560 | 0.000000 | 0.000000 | 0.00000 | 0.000000 | 0.00000 |
2 rows × 78 columns
[40]:
circoplot_df2 = spot_cellfreq_df.drop(columns=['sample'])
# Group by 'Condition' and calculate the mean of the other columns
circoplot_df2 = circoplot_df2.groupby('Condition').mean().reset_index()
circoplot_df2 = circoplot_df2.set_index('Condition')
circoplot_df2
[40]:
| Antibody.secreting.B.cells | CD3+.alpha.beta.T.cells | Central.venous.LSECs | Cholangiocytes | Erthyroid.cells | Hep | Hep.1 | Hep.3 | Hep.4 | Hep.5 | ... | Inflammatory.macrophages | Mature.B.cells | NK.like.cells | Non.inflammatory.macrophages | Periportal.LSECs | Portal.endothelial.cells | Stellate.cells | gamma.delta.T.cells.1 | tumor_1 | tumor_2 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Condition | |||||||||||||||||||||
| HCC | 0.014845 | 0.131641 | 0.009099 | 0.007342 | 0.00000 | 0.01286 | 0.000000 | 0.000000 | 0.000000 | 0.000000 | ... | 0.114776 | 0.035472 | 0.002622 | 0.034992 | 0.064790 | 0.005422 | 0.089846 | 0.003930 | 0.348701 | 0.12329 |
| Healthy | 0.019455 | 0.088204 | 0.028539 | 0.042334 | 0.00199 | 0.00000 | 0.095709 | 0.101915 | 0.185243 | 0.116906 | ... | 0.042112 | 0.019463 | 0.014303 | 0.021344 | 0.077415 | 0.007962 | 0.111953 | 0.006503 | 0.000000 | 0.00000 |
2 rows × 21 columns
[42]:
patient_group = 'HCC'
eco.create_circos_plot(circoplot_df1.loc[[patient_group]],
cell_type_colors_hex=None,
cell_abundance=circoplot_df2.loc[[patient_group]],
threshold=0.05,
edge_weights_scaler=10,
highlighted_edges=highlighted_comb,
node_weights_scaler=5000,
figure_size=(8,8),
save_path='/Users/Emrys/Downloads/'+patient_group+'_hotspots_circoplot.svg')
[ ]: