def main(save_folder, adata):
tissue_layers = ['upper EPIDERMIS', 'middle EPIDERMIS', 'basal EPIDERMIS']
img_key = 'hires'
# 1. Get cytokines and responders
t_cell_cytocines, cyto_resps_list, cytokine_responders = gene_lists.get_publication_cyto_resps(
)
# 3. Add meta data like which samples belong to which donor (optional)
if "patient" not in adata.obs_keys():
adata, tissue_cell_labels, disease_labels, lesion_labels = ctools.add_metadata(
adata)
# 1.2 Remove spots having no tissue/cell labels (since 06.10.2020)
adata = adata[np.where(
adata.obs[tissue_cell_labels].to_numpy().any(axis=1))[0]]
# 4. Use only tissue tyoes of interest
# 4.1 Add tissue types
adata = ctools.add_tissue_obs(adata)
# 4.2 Subset adata to tissue_types of interest: upper EPIDERMIS', 'middle EPIDERMIS', 'basal EPIDERMIS'
bool_col = adata.obs[tissue_layers] == 1
merged = bool_col.sum(axis=1)
adata = adata[merged == 1]
# Rename tissue region 'INTERFACE' to basal EPIDERMIS because some spots got both labels
m_interface = adata.obs['tissue_type'] == 'INTERFACE'
adata.obs['tissue_type'][m_interface] = 'basal EPIDERMIS'
"""Paper Figure 4D: Highlight cytokine and responder genes containing spots and UMI-counts in the EPIDERMIS """
convert_categories_cytokines_responders_others(
adata,
cyto_responder_genes=cytokine_responders,
save_folder=save_folder,
img_key=img_key)
def main(save_folder, spatial_adata):
"""Read out data for ST DGE Analysis and create UMAPs for Figure 3A
:return:
"""
spatial_cluster_label = 'tissue_type'
# load data
cytokines, allinone, cytoresps_dict = gene_lists.get_publication_cyto_resps(
)
# remove all spots without a tissue label
spatial_adata = spatial_adata[
spatial_adata.obs[spatial_cluster_label] != 'Unknown']
# 1. get observable for cytokine genes
spatial_adata, obs_name = add_observables.convert_variable_to_observable(
adata=spatial_adata,
gene_names=cytokines,
task='cell_gene',
label='celltype',
condition=None)
# 2. Highlight tissues epidermis and dermis + cytokines and for each single cytokine
plot_tissuerlayers_cyto(adata=spatial_adata,
obs_name='cytokine_IL17A',
title='Wholedataset_IL17A',
save_folder=save_folder,
regions=spatial_cluster_label)
def main(save_folder, spatial_adata):
"""Read out data for ST and scRNA-seq DGE Analysis and create UMAPs for Figure 3A/E and Suppl. Figures 3
:return:
"""
spatial_cluster_label = 'tissue_type'
# load data
cytokines, allinone, cytoresps_dict = gene_lists.get_publication_cyto_resps(
)
leukocyte_markers = gene_lists.leukocyte_markers()
# remove all spots without a tissue label
spatial_adata = spatial_adata[
spatial_adata.obs[spatial_cluster_label] != 'Unknown']
# 1. get observable for cytokine genes
spatial_adata, obs_name = add_observables.convert_variable_to_observable(
adata=spatial_adata,
gene_names=cytokines,
task='cell_gene',
label='celltype',
condition=None)
spatial_adata, _ = add_observables.convert_variable_to_observable(
adata=spatial_adata,
gene_names=leukocyte_markers,
task='cell_gene',
label='celltype',
condition=None)
# # 2. Read out counts and metaData for DGE Analysis including double positive cytokine cells
# 2.1 Read out only leukocytes spots by 'CD2', 'CD3D', 'CD3E', 'CD3G', 'CD247' and 'PTPRC' surface markers
adata_leukocytes = get_celltypes_data(spatial_adata,
genes=leukocyte_markers)
# 2.2 Merge layers of epidermis and save it as epidermis and merge dermis depths and save it as dermis
adata_leukocytes = get_tissueregions(adata=adata_leukocytes,
tissue_label=spatial_cluster_label)
# 3. Highlicht tissues epidermis and dermis + cytokines and for each single cytokine
plot_tissueregions_cyto(adata=adata_leukocytes,
obs_name='cytokine_IL13',
title='Leukocytes_IL13',
save_folder=save_folder)
plot_tissueregions_cyto(adata=adata_leukocytes,
obs_name='cytokine_IFNG',
title='Leukocytes_IFNG',
save_folder=save_folder)
# 4. Read out all leukocyte positive spots
include_cytokine_dp(adata=adata_leukocytes,
cytokines=cytokines,
save_folder=save_folder,
label=spatial_cluster_label,
key='ST',
paper_figure='3AC_Leukocytes')
def main(save_folder, adata):
"""Read out spots for DGE analysis and create UMAP of single cell RNA-seq data for Suppl. Figures 4B
Parameters
----------
save_folder : str
adata : annData
Returns
-------
"""
sc_cluster_obs = 'cluster_labels'
# 1. load gene list
cytokines, allinone, cytoresps_dict = gene_lists.get_publication_cyto_resps(
)
leukocyte_markers = gene_lists.leukocyte_markers()
# 2. get observable for cytokine genes
adata, obs_name = add_observables.convert_variable_to_observable(
adata=adata,
gene_names=cytokines,
task='cell_gene',
label='celltype',
condition=None)
# Only Leukocytes:
# 3. Read out counts and metaData for DGE Analysis including double positive cytokine cells
# 3.1 Read out only T-cell spots by CD2 surface markers
adata_leukocytes = get_celltypes_data(adata, genes=leukocyte_markers)
# 3.3 Read out all leukocyte cells
include_cytokine_dp(adata=adata_leukocytes,
cytokines=cytokines,
save_folder=save_folder,
label=sc_cluster_obs,
key='SC_merged',
paper_figure='SC')
# 3.4 Add IL17A label to adata
adata_leukocytes = add_observables.add_columns_genes(
adata=adata_leukocytes, genes='IL17A', label='IL17A')
""" Figure 3D: Highlight IL-17A """
plot_annotated_cells(adata=adata_leukocytes,
color='IL17A_label',
paper_figure='D',
save_folder=save_folder,
key='SC',
title='Leukocytes_IL17A',
xpos=0.02,
ypos=0.95)
def main(save_folder, adata):
"""Read out spots for DGE analysis and create UMAP of single cell RNA-seq data for Suppl. Figures 4B
Parameters
----------
save_folder : str
adata : annData
Returns
-------
"""
sc_cluster_obs = 'cluster_labels'
# 1. load gene list
cytokines, allinone, cytoresps_dict = gene_lists.get_publication_cyto_resps()
leukocyte_markers = gene_lists.leukocyte_markers()
# 2. get observable for cytokine genes
adata, obs_name = add_observables.convert_variable_to_observable(
adata=adata, gene_names=cytokines, task='cell_gene', label='celltype', condition=None)
# Only Leukocytes:
# 3. Read out counts and metaData for DGE Analysis including double positive cytokine cells
# 3.1 Read out only T-cell spots by CD2 surface markers
adata_leukocytes = get_celltypes_data(adata, genes=leukocyte_markers)
# 3.3 Read out all leukocyte cells
include_cytokine_dp(adata=adata_leukocytes, cytokines=cytokines, save_folder=save_folder,
label=sc_cluster_obs, key='SC_merged', paper_figure='SC')
# 3.3 Read out all leukocyte cells but exclude double positive cytokine cells
adata_leukocytes, obs_name = exclude_cytokine_dp(adata=adata_leukocytes, cytoresps_dict=cytoresps_dict)
# Plot cytokines and highlight double positive
plot_annotated_cells(adata=adata_leukocytes, color='cytokines_others', paper_figure='',
save_folder=save_folder, key='SC', title="Leukocytes_IL17A_IFNG",
xpos=0.02, ypos=0.95, palette=["#ff7f00", "#377eb8", 'purple'])
# Add cytokine label to adata and Plot: Highlight cytokines
adata_leukocytes = add_observables.add_columns_genes(adata=adata_leukocytes, genes='IFNG', label='IFNG')
""" Suppl. Figure 4B: Highlight IFN-g """
plot_annotated_cells(adata=adata_leukocytes, color='IFNG_label', paper_figure='4B', save_folder=save_folder,
key='SC', title="Leukocyte_IFNG", xpos=0.02, ypos=0.95)
def main(save_folder, adata):
img_key = 'hires'
# 1. Get cytokines and responders
t_cell_cytocines, cyto_resps_list, cytokine_responders = gene_lists.get_publication_cyto_resps(
)
# 3. Add meta data like which samples belong to which donor (optional)
if "patient" not in adata.obs_keys():
adata, tissue_cell_labels, disease_labels, lesion_labels = ctools.add_metadata(
adata)
# 1.2 Remove spots having no tissue/cell labels (since 06.10.2020)
adata = adata[np.where(
adata.obs[tissue_cell_labels].to_numpy().any(axis=1))[0]]
"""Paper Figure 4B: Highlight cytokine and responder genes containing spots and UMI-counts """
convert_categories_cytokines_responders_others(
adata,
cyto_responder_genes=cytokine_responders,
save_folder=save_folder,
img_key=img_key)
def main(save_folder, pp_adata, cluster_algorithm):
"""
1. scRNAseq data set
Read ou pre-processed Count matrix and apply Leiden clustering with resolution r = 0.1 for scRNAseq data set
Annotate clusters manually
2. ST data set
Visualise count matrix with tissue types
:return:
"""
# 1. load gene list
cytokines, allinone, cytoresps_dict = gene_lists.get_publication_cyto_resps(
)
# 2. Get observable for cytokine genes
pp_adata, _ = add_observables.convert_variable_to_observable(
adata=pp_adata,
gene_names=cytokines,
task='cell_gene',
label='celltype',
condition=None)
# 3. Apply cluster algorithm
pp_adata, key = apply_clusteralgo(adata=pp_adata,
algorithm=cluster_algorithm,
resolution=0.1)
# 4. Annotate clusters with expert opinion - before the best resolution r=0.1 was identified
pp_adata = annotate_cluster(adata=pp_adata,
cluster_algorithm=cluster_algorithm,
resolution=0.1)
# 5. Plot UMAP scRNAseq data
visualise_clusters(adata=pp_adata,
save_folder=save_folder,
key='cluster_labels',
title="SC")
def main(adata, save_folder, tissue_types, radii, get_plots=False):
"""
Parameters
----------
adata : annData
save_folder : str
tissue_types : str, list
radii : int, list
get_plots : bool
Returns
-------
"""
# 1. Get cytokines and responders
conditional_genes, _, conditionalgenes_responders = gene_lists.get_publication_cyto_resps()
# 2. prepare adata object
adata = data_preparation(adata=adata, tissue_types=tissue_types, conditional_genes=conditional_genes,
conditionalgenes_responders=conditionalgenes_responders)
# 3. Run conditional clustering and calculate (spatially weighted) correlation
sig = []
if isinstance(radii, list):
for radius in radii:
sig = run_spatialcorrelation(adata=adata, tissue_types=tissue_types,
cytokine_responders=conditionalgenes_responders, save_folder=save_folder,
radius=radius, sig=sig, get_plots=get_plots)
else:
sig = run_spatialcorrelation(adata=adata, tissue_types=tissue_types,
cytokine_responders=conditionalgenes_responders,
save_folder=save_folder, radius=radii, sig=sig, get_plots=get_plots)
if len(sig) > 1:
# 6. Evaluate distance via elbow plot
plot_evaluations.plot_evaluate_distance(
significance=sig, cytokines=conditional_genes, save_folder=save_folder)
def main(dataset_type, save_folder, df_keys, log, dge_results_folder):
"""
Parameters
----------
dataset_type : str
save_folder : str
df_keys : list
log : bool
dge_results_folder : str
Returns
-------
"""
# Determine name of cluster observable
if dataset == 'SC':
cluster_label = 'cluster_labels'
else:
cluster_label = 'tissue_type'
print("# ------ Load data ------ #")
cytokines, allinone, cytoresps_dict = gene_lists.get_publication_cyto_resps(
)
genes_to_highlight = gene_lists.highlight_genes()
# # 1. Load adata
adata = load_adata(type_dataset=dataset_type, cluster_label=cluster_label)
# # 2 Read out only T-cell spots by leukocyte markers
adata = get_sub_adata(adata=adata, gene=gene_lists.leukocyte_markers())
# # 3. Assign condition to spots
adata, observable = exclude_cytokine_dp(adata=adata,
cytoresps_dict=cytoresps_dict)
for cyto in cytokines:
# get observable of condition
obs_label_condition = "_".join(['cytokine', cyto])
genes_labeling = genes_to_highlight[cyto]
all_csv_files = [
file for path, subdir, files in os.walk(dge_results_folder)
for file in glob.glob(os.path.join(path, '*.csv'))
]
pattern = "".join(['*', cyto, '*all_genes.csv'])
for file in all_csv_files:
if fnmatch(file, pattern):
# Read out only those driver and responder genes which are specific for a cytokine
allgenes_df = pd.read_csv(file, error_bad_lines=False)
# Remove column Unnamed: 0
allgenes_df = allgenes_df.drop(['Unnamed: 0'], axis=1)
# Check if column names and row names are unique
print("Unique Genes Stratified Sampling:",
allgenes_df['gene_symbol'].is_unique)
print("Unique Genes Stratified Sampling:",
allgenes_df.columns.is_unique)
# remove duplicated rows
allgenes_df = allgenes_df.loc[
~allgenes_df['gene_symbol'].duplicated(), :]
# Name of used design function
design = file.split(os.sep)[-4]
# Name of used DGE Analysis method
method = file.split(os.sep)[-1].split("_")[-4]
# Create output folder
output_folder = os.path.join(save_folder, design, cyto)
os.makedirs(output_folder, exist_ok=True)
allgenes_df = _write_dataframe(adata,
df=allgenes_df,
cytokine=cyto,
observable=obs_label_condition,
output_folder=output_folder,
method=method)
print("# ------ Volcano plot ------ #")
# 3. Volcano plot interactive plot
plotly_interactive_volcano(
df=allgenes_df,
df_keys=df_keys,
save_folder=output_folder,
key="".join([method, "_", cyto, "+", "_vs_", cyto, "-"]),
x_lab=r'log$_2$(FC)',
y_lab=r'-log$_{10}$(pvalue)',
log2fc_cut=1,
pval_cut=0.05)
volcano_plot(df=allgenes_df,
df_keys=df_keys,
cytokine=cyto,
adjust=True,
label_genes=genes_labeling,
title="_".join(
[method, cyto, "Volcano_plot_zoom"]),
save_folder=output_folder,
log2fc_cut=1.0,
threshold=0.05)
print(
"# ------ Violin plots of Novel and Golden Standard genes ------ #"
)
# Get expression level of genes of interest to create boxplots
if isinstance(genes_labeling, dict):
driver_group = genes_to_highlight[cyto]['Driver_genes']
responder_group = genes_to_highlight[cyto][
'Responder_genes']
merged_genes = driver_group.copy()
merged_genes.extend(responder_group)
else:
merged_genes = genes_labeling.copy()
# First check if genes are in data set
available_genes = list(
set(adata.var.index) & set(merged_genes))
available_genes.append(cyto)
# 3. get counts for each gene of interest
goi = pd.DataFrame()
goi['Cyto+_vs_Cyto-'] = adata.obs[obs_label_condition].values
# available_genes = ['IL17A', 'IFNG', 'IL13']
for gene in available_genes:
if gene in adata.var_names:
# Get counts
adata_gene, new_obs_name = get_expression_values(
adata=adata, gene=gene)
# 4. sub-divide into cyto+ and cyto- group and read out counts
spatial_adata_gene = create_obs_cytopos_cytoneg(
adata=adata_gene,
cyto=cyto,
gene=gene,
observable=obs_label_condition)
goi[gene] = spatial_adata_gene.obs["_".join(
[gene, 'group'])].values
# 5. Visualise counts of gene of interest in a violin plot
plot_violins(adata=spatial_adata_gene,
group=gene,
groupby=obs_label_condition,
output_folder=output_folder,
log=log)
# Save counts of genes of interest
goi.to_csv(
os.path.join(
output_folder,
"_".join([cyto, "Counts_Highlight_genes.csv"])))
def main(save_folder, spatial_adata):
"""
Read out data for ST and scRNA-seq DGE Analysis and create UMAPs for Figure 3A/E and Suppl. Figures 3
:return:
"""
spatial_cluster_label = 'tissue_type'
# 1. load gene lists
cytokines, allinone, cytoresps_dict = gene_lists.get_publication_cyto_resps(
)
leukocyte_markers = gene_lists.leukocyte_markers()
# 2. remove all spots without a tissue label
spatial_adata = spatial_adata[
spatial_adata.obs[spatial_cluster_label] != 'Unknown']
# 3. get observable for cytokine genes and leukocyte markers
spatial_adata, obs_name = add_observables.convert_variable_to_observable(
adata=spatial_adata,
gene_names=cytokines,
task='cell_gene',
label='celltype',
condition=None)
spatial_adata, obs_name = add_observables.convert_variable_to_observable(
adata=spatial_adata,
gene_names=leukocyte_markers,
task='cell_gene',
label='celltype',
condition=None)
# 4. Read out only leukocytes spots by 'CD2', 'CD3D', 'CD3E', 'CD3G', 'CD247' and 'PTPRC' surface markers
adata_leukocytes = get_celltypes_data(spatial_adata,
genes=leukocyte_markers)
# 5. add observable healthy_disease
spatial_adata = add_observables.add_disease_healthy_obs(spatial_adata)
# keys: 'patient', 'biopsy_type', 'disease', 'tissue_type'
# Suppl Figure 2A
visualise_clusters(adata=spatial_adata,
save_folder=save_folder,
key='healthy_disease',
title="Diagnoses")
# Suppl. Figure 2C
visualise_clusters(adata=adata_leukocytes,
save_folder=save_folder,
key='tissue_type',
title="Leukocytes_tissuelayers")
# 5. Read out spots which either have IL17A, IL13 or INFG genes
adata_leukocytes, obs_name = exclude_cytokine_dp(
adata=adata_leukocytes, cytoresps_dict=cytoresps_dict)
# 6. Merge layers of epidermis and save it as epidermis and merge dermis depths and save it as dermis
adata_leukocytes = get_tissueregions(adata=adata_leukocytes,
tissue_label=spatial_cluster_label)
# Suppl. Figure 2B
plot_tissueregions_cyto(adata=adata_leukocytes,
obs_name=obs_name,
title='Leukocytes_Cytokines',
save_folder=save_folder,
gene_colors=[
"#ff7f00", "#e41a1c", 'darkgoldenrod',
'purple', "#377eb8", 'deeppink'
])
