This page was generated from a Jupyter notebook. You can download the notebook here to run it locally in Jupyter.

Ductal carcinoma in situ

Thor performs unbiased screening of breast cancer hallmarks in Ductal Carcinoma in situ sample and reveals heterogeneity of immune response.

Run Thor

In this tutorial, we show inference of cell-level spatial gene expression from the Visium spot-level spatial transcriptome and a whole slide image (H&E staining) using Thor.

The spatial dataset is Human Breast Cancer: Ductal Carcinoma In Situ, Invasive Carcinoma (FFPE). The input data are downloaded from 10x Genomics website.

Thor processed data can be downloaded directly from google drive.

For installation of Thor, please refer to this installation guide.

Import the packages

import sys
import os
import logging
import datetime

logger = logging.getLogger()
logger.setLevel(logging.INFO)
logging.basicConfig(format='%(name)s - %(levelname)s - %(message)s')

now = datetime.datetime.now()
logger.info(f"Current Time: {now}")

root - INFO - Current Time: 2023-12-13 16:53:16.118962

%config InlineBackend.figure_format = 'retina'


import numpy as np
import pandas as pd
import scanpy as sc
import seaborn as sns
import matplotlib.pyplot as plt
sc.set_figure_params(scanpy=True, dpi=80, dpi_save=300)
sc.settings.verbosity = 'error'


from thor.pp import WholeSlideImage, Spatial
from thor.finest import fineST

numexpr.utils - INFO - Note: NumExpr detected 12 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 8.
numexpr.utils - INFO - NumExpr defaulting to 8 threads.

Preprocessing

Preprocessing the whole slide image, including cell segmentation and morphological feature extraction. If not specified, the segmentation data will be saved in a directory (created by Thor) WSI_{sn}.

The WSI can be downloaded directly from google drive.

The outcomes are: cell mask and the cell feature.csv

sn = 'DCIS_10x'
image_path = "Visium_FFPE_Human_Breast_Cancer_image.tif"

wsi = WholeSlideImage(image_path, name=sn)

Here we use Cellpose to segment the nuclei. Notice the segmentation can take a long time without GPU. The output file is a n_pixel_row x n_pixel_col matrix. We’ll skip this step and load the pre-computed segmentation results from here.

%%script echo "Skip cell segmentation"

wsi.process(method='cellpose')

Skip cell segmentation

Extract handcrafted image features based on the image and nuclei segmentation result.

cell_mask_path = os.path.join('WSI_DCIS_10x', 'cell_mask.npz')
wsi = WholeSlideImage(image_path, name=sn, nuclei_seg_path=cell_mask_path, nuclei_seg_format='mask_array_npz')
wsi.process()

[12/13/23 16:53:37] INFO     Thor: 76396 cells detected.
                    INFO     Thor: 76396 cells mapped to spots.
[12/13/23 16:53:38] INFO     Thor: 76359 cells kept.
<function mean at 0x107c40900>
                    INFO     Thor: Extracting image features from a square image patch of size 112 pixels centered at the cell.

Extracting gray image features: 100%|██████████| 76359/76359 [00:04<00:00, 16574.04it/s]
Extracting RGB features: 100%|██████████| 76359/76359 [00:33<00:00, 2297.83it/s]

Preprocessing spatial transcriptome. We used SCANPY to generate an adata for the spots including QC.

We follow the standarized steps used by SCANPY to create the spot adata from the space ranger output directory (visium_dir). The spot adata contains the expression matrix, the location of the spots as pixel positions on the WSI, and the hires, lowres images with scalefactors. - spot.h5ad

We skip this step here and the files can be downloaded from here

Predicting cell-level gene expression using Thor diffusion

After finishing the preprocessing, you should have those files: - The original WSI (image_path) - The cell(nuclei) mask and features (in directory “./WSI_DCIS_10x”) - The spot-level gene expression (in directory “./Spatial_DCIS_10x”)

outdir = os.getcwd()

image_process_dir = os.path.join(outdir, "WSI_DCIS_10x")
cell_mask_path = os.path.join(image_process_dir, "cell_mask.npz")
cell_feature_path = os.path.join(image_process_dir, "cell_features.csv")

spatial_dir = os.path.join(outdir, "Spatial_DCIS_10x")
spot_adata_path = os.path.join(spatial_dir, "spot.h5ad")

The first step is to map the spot gene expression to the segmented cells. We use the nearest neighbors approach. This cell-level gene expression is the starting point for the diffusion process.

DCIS = fineST(
    image_path,
    name="DCIS_10x",
    spot_adata_path=spot_adata_path,
    cell_features_csv_path=cell_feature_path
)

DCIS.prepare_input(mapping_margin=10)

[12/13/23 16:54:25] INFO     Thor: Please check alignment of cells and spots
                    INFO     Thor: The first two columns in the node_feat DataFrame need to match the spatial coordinates from obsm['spatial'].
                    INFO     Thor: Mapping cells to the closest spots within 10 x the spot radius

DCIS.adata

AnnData object with n_obs × n_vars = 76344 × 16784
    obs: 'in_tissue', 'array_row', 'array_col', 'n_genes', 'n_genes_by_counts', 'total_counts', 'total_counts_mt', 'pct_counts_mt', 'spot_barcodes', 'x', 'y', 'mean_gray', 'std_gray', 'entropy_img', 'mean_r', 'mean_g', 'mean_b', 'std_r', 'std_g', 'std_b', 'spot_heterogeneity'
    var: 'gene_ids', 'feature_types', 'genome', 'n_cells', 'mt', 'n_cells_by_counts', 'mean_counts', 'pct_dropout_by_counts', 'total_counts'
    uns: 'spatial', 'cell_image_props'
    obsm: 'spatial'

There are 75117 cells and for the sake of time, we’ll show prediction of a few genes. The same Markov transition matrix can be applied to all genes. The user-defined genes can be input by either in a 1-column text file or directly as an attribute of the DCIS object

DCIS.genes = ['VEGFA', 'FGFR4', 'TPD52', 'GRB7', 'JUP', 'SCGB2A2', 'KANK1', 'ESR1', 'TFRC', 'ERBB2']

DCIS.set_genes_for_prediction(genes_selection_key=None)

DCIS.adata.var[DCIS.adata.var['used_for_prediction']]

	gene_ids	feature_types	genome	n_cells	mt	n_cells_by_counts	mean_counts	pct_dropout_by_counts	total_counts	used_for_prediction
TFRC	ENSG00000072274	Gene Expression	GRCh38	1847	False	1847	2.779190	26.648133	6998.0	True
FGFR4	ENSG00000160867	Gene Expression	GRCh38	165	False	165	0.094122	93.447180	237.0	True
VEGFA	ENSG00000112715	Gene Expression	GRCh38	1900	False	1900	3.851866	24.543288	9699.0	True
ESR1	ENSG00000091831	Gene Expression	GRCh38	166	False	166	0.069102	93.407466	174.0	True
TPD52	ENSG00000076554	Gene Expression	GRCh38	2016	False	2016	5.621922	19.936458	14156.0	True
KANK1	ENSG00000107104	Gene Expression	GRCh38	818	False	818	0.499603	67.513900	1258.0	True
SCGB2A2	ENSG00000110484	Gene Expression	GRCh38	2213	False	2213	17.403097	12.112788	43821.0	True
ERBB2	ENSG00000141736	Gene Expression	GRCh38	2417	False	2417	22.986498	4.011120	57880.0	True
GRB7	ENSG00000141738	Gene Expression	GRCh38	1411	False	1411	2.008737	43.963463	5058.0	True
JUP	ENSG00000173801	Gene Expression	GRCh38	1758	False	1758	2.829230	30.182685	7124.0	True

DCIS.recipe = 'gene'
DCIS.set_params(
    is_rawCount=False,
    out_prefix="fineST",
    write_freq=10,
    n_iter=20,
    conn_csr_matrix="force",
    smoothing_scale=0.8,
    node_features_obs_list=['spot_heterogeneity'],
    n_neighbors=5,
    geom_morph_ratio=1,
    geom_constraint=0,
    inflation_percentage=None,
    regulate_expression_mean=False,
    stochastic_expression_neighbors_level='spot',
    smooth_predicted_expression_steps=1,
    reduced_dimension_transcriptome_obsm_key=None,
    adjust_cell_network_by_transcriptome_scale=0,
    n_jobs=4)

DCIS.predict_gene_expression()

[12/13/23 16:54:27] INFO     Thor: Using mode gene
                    INFO     Thor: Forcing to recalculate the connectivities.
                    INFO     Thor: Construct SNN with morphological features: ['mean_gray', 'std_gray', 'entropy_img', 'mean_r', 'mean_g', 'mean_b',
                             'std_r', 'std_g', 'std_b'].
[12/13/23 16:54:28] INFO     Thor: Finish constructing SNN
                    INFO     Thor: Add adata.obsp["snn_connectivities"]
                    INFO     Thor: Add adata.obsp["snn_knn_connectivities"]
                    INFO     Thor: Add adata.uns["snn"]
                    INFO     Thor: Weigh cells according to the spot heterogeneity.
[12/13/23 16:54:29] INFO     Thor: Promote flow of information from more homogeneous cells to less.
                    INFO     Thor: Eliminate low quality edges (<0.1) between cells.
                    INFO     Thor: Compute transition matrix
                    INFO     Thor: self weight scale is set to: 0.200
                    INFO     Thor: Added transition matrix to adata.obsp["snn_transition_matrix"]

... storing 'spot_barcodes' as categorical

[12/13/23 16:54:30] INFO     Thor: fineST estimation starts.
[12/13/23 16:54:48] INFO     Thor: fineST estimation finished.
                    
                    

Check the expression of a few genes.

ad_thor = DCIS.load_result('fineST_20_samp1.npz')
ad_thor

AnnData object with n_obs × n_vars = 76344 × 10
    obs: 'in_tissue', 'array_row', 'array_col', 'n_genes', 'n_genes_by_counts', 'total_counts', 'total_counts_mt', 'pct_counts_mt', 'spot_barcodes', 'x', 'y', 'mean_gray', 'std_gray', 'entropy_img', 'mean_r', 'mean_g', 'mean_b', 'std_r', 'std_g', 'std_b', 'spot_heterogeneity', 'node_weights'
    var: 'gene_ids', 'feature_types', 'genome', 'n_cells', 'mt', 'n_cells_by_counts', 'mean_counts', 'pct_dropout_by_counts', 'total_counts', 'used_for_prediction', 'used_for_reduced', 'used_for_vae'
    uns: 'spatial', 'cell_image_props', 'snn'
    obsm: 'spatial'
    obsp: 'snn_connectivities', 'snn_knn_connectivities', 'snn_transition_matrix'

genes = DCIS.genes

ad_spot = sc.read_h5ad(spot_adata_path)
sc.pl.spatial(ad_spot, color=genes, vmax='p99', frameon=False)

genes = DCIS.genes
sc.pl.spatial(ad_thor, color=genes, spot_size=50, vmax='p99', frameon=False)

Advance analysis

In this notebook, we’ll perform analyses on the Ductal Carcinoma In Situ (DCIS) spatial data at cell-level inferred by Thor, as follows,

• Visualization of genes in the physical context at cell level

• Tumor activity using oncogenes and tumor suppressor genes

• Hallmark pathway enrichment

• Copy number variation calculation

• Tertiary Lymphoid Structures quantification

For installation of Thor, please refer to Thor website.

The cell-level expression of the highly variable genes (2747 genes) inferred by Thor can be downloaded from the link. You can also download the nuclei segmentation masks and full resolution image in the same link if you skip first part of the notebooks in this case study.

Import packages

import os
import scanpy as sc
import numpy as np
from scipy import stats
import seaborn as sns
import matplotlib as mpl
from matplotlib import pyplot as plt
from PIL import Image
from shapely.geometry import Polygon


import thor
from thor.utils import convert_pixel_to_micron_visium, get_ROI_tuple_from_polygon, get_adata_layer_array
from thor.pl import annotate_ROI, get_nuclei_pixels, single_molecule
from thor.analy import get_pathway_score

%config InlineBackend.figure_format = 'retina'
sc.set_figure_params(scanpy=True, dpi=80, dpi_save=300)
sc.settings.verbosity = "error"

Load the image (for visualization) and Thor-predicted cell-level transcriptome result

image_path = "Visium_FFPE_Human_Breast_Cancer_image.tif"
fullres_im = np.array(Image.open(image_path))

adata_path = os.path.join("Thor_DCIS_10x", "cell_data_thor.h5ad")
ad = sc.read_h5ad(adata_path)
ad

AnnData object with n_obs × n_vars = 76320 × 2748
    obs: 'in_tissue', 'array_row', 'array_col', 'n_genes', 'n_genes_by_counts', 'total_counts', 'total_counts_mt', 'pct_counts_mt', 'spot_barcodes', 'x', 'y', 'mean_gray', 'std_gray', 'entropy_img', 'mean_r', 'mean_g', 'mean_b', 'std_r', 'std_g', 'std_b', 'spot_heterogeneity', 'node_weights', 'tumor_region', 'tumor_region_ext', 'copy_variation'
    var: 'gene_ids', 'feature_types', 'genome', 'n_cells', 'mt', 'n_cells_by_counts', 'mean_counts', 'pct_dropout_by_counts', 'total_counts', 'used_for_prediction', 'used_for_reduced', 'used_for_vae'
    uns: 'cell_image_props', 'copy_variation_colors', 'hvg', 'leiden', 'leiden_colors', 'neighbors', 'pca', 'recluster_normal_colors', 'snn', 'spatial', 'tumor_cluster_CNV_colors', 'tumor_normal_subtumor_colors', 'tumor_region_boundaries', 'umap'
    obsm: 'spatial'
    obsp: 'snn_connectivities', 'snn_knn_connectivities', 'snn_transition_matrix'

Visualize genes at tissue and cell level

# Show all the tumor regions
# '0' means non-tumor
sc.pl.spatial(ad, color='tumor_region', spot_size=50, cmap='tab20b', title='all major tumor regions')

palette = thor.pl.colors.continuous_palettes['blueYellow']
cmap = mpl.colors.LinearSegmentedColormap.from_list('my_cmap', palette, N=256)

gene = 'VEGFA'
sc.pl.spatial(ad, color='VEGFA', spot_size=50, vmax='p99', cmap=cmap)

Close-up gene expression in cells can be seamlessly explored using our dedicated webapp Mjolnir or as static figures plotted using Thor’s plotting functions. Here we annotate tumor regions on the H & E image according to Agoko NV, Belgium on 10x Genomics website.

We then use Mjolnir to extract the tumor regions and gene expression matrices of the cells in those regions.

tumor1 = ad.uns['tumor_region_boundaries']['1']
tumor1_polygon = Polygon(tumor1)
ROI_tuple = get_ROI_tuple_from_polygon(tumor1, extend_pixels=100)
annotate_ROI(fullres_im, ROI_polygon=tumor1_polygon, lw=4)

load the cellmasks

cells_pixels = get_nuclei_pixels(ad, "WSI_DCIS_10x/cell_mask.npz")
microPerPixel = convert_pixel_to_micron_visium(ad)

gene = 'VEGFA'
expression_vector = get_adata_layer_array(ad[:, gene])

ax = single_molecule(gene, expression_vector, cells_pixels, full_res_im=fullres_im, ROI_tuple=ROI_tuple, cmap=cmap, img_alpha=0.5, vmax='p99', dpi=500, figsize=(2,2))
thor.plotting.fill.add_scalebar(ax, microPerPixel, 'um', 'lower left')

ROI_squares = [[9000, 7900, 500, 500], [10400, 8900, 500, 500]]

fig, axes = plt.subplots(2, 2)

i = 1
for r in ROI_squares:
    single_molecule(gene, expression_vector, cells_pixels, full_res_im=fullres_im, ROI_tuple=r, img_alpha=1, figsize=(3, 3), dpi=200, alpha=0, show_cbar=False, return_fig=False, show=False, cmap=cmap, ax=axes[i, 0])
    ax=single_molecule(gene, expression_vector, cells_pixels, full_res_im=fullres_im, ROI_tuple=r, img_alpha=0.5, figsize=(3, 3), dpi=200, alpha=1, show_cbar=False, return_fig=False, show=False, cmap=cmap, vmax='p99', ax=axes[i, 1])
    thor.plotting.fill.add_scalebar(ax, microPerPixel, 'um', 'lower left')
    i = i - 1

Expression profile of oncogene and tumor suppressor gene

sc.set_figure_params(dpi=80, vector_friendly=True, fontsize=10, frameon=False, figsize=(6,6))
sc.pl.spatial(ad, color=["ERBB2", "ATM"], spot_size=50, vmax=4, cmap='viridis', img_key="hires", show=False)

[<Axes: title={'center': 'ERBB2'}, xlabel='spatial1', ylabel='spatial2'>,
 <Axes: title={'center': 'ATM'}, xlabel='spatial1', ylabel='spatial2'>]

g1 ="ERBB2"
g2 = "ATM"
region_col = "tumor_region"
for region in ["7", "1", "17", "11", "6", "15"]:

    x = ad[ad.obs[region_col]==region, g1].X[:, 0]
    y = ad[ad.obs[region_col]==region, g2].X[:, 0]

    values = np.vstack([x, y])
    kernel = stats.gaussian_kde(values)(values)
    ax1 = sns.jointplot(x=x, y=y, kind="kde", fill=True, levels=100)

    ax1.ax_joint.cla()
    plt.sca(ax1.ax_joint)
    ax = sns.scatterplot(x=x, y=y, s=10, alpha=1, c=kernel, cmap="jet", ax=ax1.ax_joint)
    ax.set_xlabel(g1)
    ax.set_ylabel(g2)
    ax.set_xlim(0, 5)
    ax.set_ylim(0, 2)

g1 ="ERBB2"
g2 = "ATM"

x = ad[:, g1].X[:, 0]
y = ad[:, g2].X[:, 0]

values = np.vstack([x, y])
kernel = stats.gaussian_kde(values)(values)
ax1 = sns.jointplot(x=x, y=y, kind="kde", fill=True, levels=100)

ax1.ax_joint.cla()
plt.sca(ax1.ax_joint)
ax = sns.scatterplot(x=x, y=y, s=3, alpha=0.5, c=kernel, cmap="jet", ax=ax1.ax_joint)
ax.set_xlabel(g1)
ax.set_ylabel(g2)
ax.set_xlim(0, 5)
ax.set_ylim(0, 2.5)

plt.show()

g1 ="ERBB2"
g2 = "ATM"
region_col = "tumor_region"

x = ad[:, g1].X[:, 0]
y = ad[:, g2].X[:, 0]

ad.obs['severe'] = np.log2(x/(y+0.01))

# tumor_region '0' is the background (non-tumor) region
ad1 = ad[ad.obs[region_col] != '0']

sc.set_figure_params(dpi=80, color_map="RdYlGn_r", vector_friendly=True, fontsize=18, frameon=False, figsize=(8,8))
sc.pl.spatial(ad1, color=["severe"], spot_size=40, vmin=-3, vmax=3, alpha_img=0.5, title="")

Hallmark pathway enrichment

Pathway enrichment analyses are done using decoupler-py.

import decoupler as dc

# Load MSigDB. This can take a while depending on the internet connection (normally 1-2 minutes)
msigdb = dc.get_resource('MSigDB')

# Filter by hallmark
msigdb = msigdb[msigdb['collection']=='hallmark']
msigdb = msigdb[~msigdb.duplicated(['geneset', 'genesymbol'])]

acts = get_pathway_score(ad, layer=None, net_df=msigdb)

sc.pp.scale(acts, max_value=10)

root - INFO - Downloading data from `https://omnipathdb.org/queries/enzsub?format=json`
root - INFO - Downloading data from `https://omnipathdb.org/queries/interactions?format=json`
root - INFO - Downloading data from `https://omnipathdb.org/queries/complexes?format=json`
root - INFO - Downloading data from `https://omnipathdb.org/queries/annotations?format=json`
root - INFO - Downloading data from `https://omnipathdb.org/queries/intercell?format=json`
root - INFO - Downloading data from `https://omnipathdb.org/about?format=text`
root - INFO - Downloading annotations for all proteins from the following resources: `['MSigDB']`

Running ora on mat with 76320 samples and 2748 targets for 49 sources.

100%|██████████| 76320/76320 [00:40<00:00, 1864.33it/s]

example_pathways = ['HALLMARK_DNA_REPAIR', 'HALLMARK_E2F_TARGETS', 'HALLMARK_ESTROGEN_RESPONSE_EARLY', 'HALLMARK_IL6_JAK_STAT3_SIGNALING']

sc.pl.spatial(acts, color=example_pathways, spot_size=50, cmap='viridis', alpha_img=0.5, ncols=2, vmax='p99', colorbar_loc=None)

Copy number variation

This requires installation of CopyKAT and R environment in order to run the script. Here we simply show our result.

from thor.analy import prepare_and_run_copykat

prepare_and_run_copykat(adata,datadir, sam_name="BC")

#sc.pl.spatial(ad, color='copy_variation', spot_size=50, ncols=1)

import seaborn as sns
import matplotlib.pyplot as plt
sc.set_figure_params(dpi=80, facecolor='white')

fig, ax = plt.subplots(figsize=(5,5))
sc.pl.spatial(ad, color='copy_variation', spot_size=50, vmax=4, cmap='viridis', img_key="hires", show=False, ax=ax)

ax.set_xticks([])
ax.set_yticks([])

plt.show()

Tertiary Lymphoid Structures(TLS) quantification

We use the list of 29 signature genes used by this study.

# This is the gene list for the TLS score that we finally used, since it includes more related cell-type markers.
TLS_list_immunity = ['IGHA1',
                    'IGHG1',
                    'IGHG2',
                    'IGHG3',
                    'IGHG4',
                    'IGHGP',
                    'IGHM',
                    'IGKC',
                    'IGLC1',
                    'IGLC2',
                    'IGLC3',
                    'JCHAIN',
                    'CD79A',
                    'FCRL5',
                    'MZB1',
                    'SSR4',
                    'XBP1',
                    'TRBC2',
                    'IL7R',
                    'CXCL12',
                    'LUM',
                    'C1QA',
                    'C7',
                    'CD52',
                    'APOE',
                    'PLTP',
                    'PTGDS',
                    'PIM2',
                    'DERL3']

sc.tl.score_genes(ad, TLS_list_immunity, ctrl_size=len(TLS_list_immunity), gene_pool=None, n_bins=25, score_name='TLS_score_immunity', random_state=0, copy=False, use_raw=False)

sc.set_figure_params(dpi=80, facecolor='white', frameon=False)


score = 'TLS_score_immunity'
ad1 = ad[ad.obs[score]>0.3]

scalefactor = ad.uns['spatial']['Visium_FFPE_Human_Breast_Cancer']['scalefactors']['tissue_hires_scalef']

fig, ax = plt.subplots(figsize=(6,6))

sc.pl.spatial(ad1, color=score, spot_size=50, cmap='viridis', ncols=1, ax=ax, show=False, title="", colorbar_loc=None)

tumor_borders = ad.uns['tumor_region_boundaries']
for name, border in tumor_borders.items():
    ax.plot(border[:,0]*scalefactor, border[:,1]*scalefactor, color='r', lw=0.5, zorder=3)


plt.show()

Rank the tumor regions according to the TLS score. We extended the tumor regions by 1 spot (55 um) outwards.

region_col = 'tumor_region_ext'
import numpy as np
scores =  ['TLS_score_immunity']

for score in scores:
    ad1 = ad[ad.obs[score]>0.]
    TLS_clone_stats1 = ad1.obs.groupby(region_col)[[score]].aggregate('median')
    rank = list(np.array(TLS_clone_stats1.sort_values(by=score, ascending=False).index))
    rank.remove('0')

    print(rank)

['7', '1', '17', '16', '4', '8', '5', '12', '9', '18', '2', '14', '10', '3', '13', '11', '6', '15']

import seaborn as sns
from matplotlib import pyplot as plt

sc.set_figure_params(dpi=80, facecolor='white')

tumor_col = 'tumor_region_ext'
score = 'TLS_score_immunity'

df = ad.obs[[region_col, score]]
df = df.loc[df[score]>0, :]
df = df.loc[df[region_col]!='0', :]

fig, ax = plt.subplots(figsize=(6, 5))

ax.spines.right.set_color('none')
ax.spines.top.set_color('none')
ax = sns.boxplot(df, y=score, x=region_col, whis=[5,95], dodge=False, width=0.3, fliersize=0., order=rank )
for patch in ax.patches:
    r, g, b, a = patch.get_facecolor()
    patch.set_facecolor((r, g, b, .5))
#sns.stripplot(df, y=score, x=region_col, ax=ax, size=0.8, jitter=0.05, dodge=True, legend="", alpha=1)
ax._remove_legend(ax.get_legend())
ax.set_ylim(0,0.6)
ax.set_yticks([0, 0.2, 0.4, 0.6])
ax.set_xlabel('Tumor region')
ax.set_ylabel('TLS score')

matplotlib.category - INFO - Using categorical units to plot a list of strings that are all parsable as floats or dates. If these strings should be plotted as numbers, cast to the appropriate data type before plotting.
matplotlib.category - INFO - Using categorical units to plot a list of strings that are all parsable as floats or dates. If these strings should be plotted as numbers, cast to the appropriate data type before plotting.

Text(0, 0.5, 'TLS score')