[ ]:
from sctoolbox.utilities import bgcolor
0B - Receptor-Ligand Analysis
1 - Description
Note: Requires notebook 04_clustering to be run prior to this notebook.
Receptor-ligand interactions play a crucial role in mediating cellular communication and signaling processes in biological systems. In the context of single-cell data analysis, studying receptor-ligand interactions provides insights into cell-cell communication networks and their impact on various cellular functions and behaviors.
This notebook provides the tools to investigate and visualize the ligand-receptor interactions between conditions and/or clusters.
2 - Setup
[ ]:
import scanpy
import sctoolbox.receptor_ligand as rl
import sctoolbox.utilities as utils
import pandas as pd
from sctoolbox import settings
utils.settings_from_config("config.yaml", key="0B")
3 - Load adata
The dataset must be clustered for a receptor-ligand analysis.
⬐ Fill in input data here ⬎
⬐ Fill in input data here ⬎
[ ]:
%bgcolor PowderBlue
anndata_file = "anndata_4.h5ad"
[ ]:
adata = utils.load_h5ad(anndata_file)
with pd.option_context("display.max.rows", 5, "display.max.columns", None):
display(adata)
display(adata.obs)
display(adata.var)
4 - General input
⬐ Fill in input data here ⬎
⬐ Fill in input data here ⬎
[ ]:
%bgcolor PowderBlue
# Path to receptor ligand database
db_path = f'http://tcm.zju.edu.cn/celltalkdb/download/processed_data/human_lr_pair.txt'
ligand_column = 'ligand_gene_symbol'
receptor_column = 'receptor_gene_symbol'
# Comput interactions
cluster_col = "clustering"
normalize = 1000 # Correct clusters to given size.
gene_col = None # Column that holds gene name /id. Set to None to use index
overwrite = False # Overwrite existing interaction table
# Plotting
dpi = 100
## Violin
violin_min_perc = 0 # Minimum percentage of cells in a cluster that express the respective gene.
violin_out_suffix = "violin" # Suffix of output file
violin_figsize = (5, 30) # Figure size
## Network
net_min_perc = 0 # Minimum percentage of cells in a cluster that express the respective gene.
net_interaction_score = 0 # Interaction score must be above this threshold for the interaction to be counted in the graph.
net_interaction_perc = 90 # Select interaction scores above or equal to the given percentile. Will overwrite parameter interaction_score.
net_out_suffix = "network" # Suffix of output file
net_title = None # Plot title
net_color_min = 0 # Min value for color range
net_color_max = None # Max value for color range
net_restrict_to = [] # Only show given clusters provided in list.
net_show_count = True # Show the interaction count
## Connection plot
con_restrict_to = None # Restrict plot to given cluster names.
con_figsize = (5, 10) # Figure size
con_out_suffix="connectionPlot" # Suffix of output file
con_title = None # Plot title
con_filter = con_filter = "receptor_score > 0 & ligand_score > 0 & interaction_score > 0 & receptor_percent >= 50 & ligand_percent >= 50" # Conditions to filter the interaction table
con_lw_multiplier = 2 # Linewidth multiplier
con_wspace = 0.4 # Width between plots
con_lig_whitelist = None # List of ligand genes that should be exclusively shown.
con_rec_whitelist = None # List of receptor genes that should be exclusively shown.
## Filter for interaction table
tab_min_perc = 0
tab_interaction_score = 0
tab_interaction_perc = 90
tab_out_suffix = "interaction_table"
group_a = None
group_b = None
Use ``help(<function_name>)`` for information on the parameters. E.g. ``help(rl.download_db)``
5 - Download receptor-ligand database
Provide a path to a table with receptor-ligand interactions. Will be stored in the adata object (adata.uns['receptor-ligand]['database']).
[ ]:
rl.download_db(adata=adata,
db_path=db_path,
ligand_column=ligand_column,
receptor_column=receptor_column,
inplace=True,
overwrite=False)
6 - Compute interactions
Create a table of receptor-ligand interactions between given groups. This table is used for plotting. Will be stored in the adata object (adata.uns['receptor-ligand']['interactions'])
[ ]:
rl.calculate_interaction_table(adata=adata,
cluster_column=cluster_col,
gene_index=gene_col,
normalize=normalize,
inplace=True,
overwrite=overwrite)
7 - Plotting
7.1 - Violin
Show the distribution of interaction scores for all group combinations.
Scores
> 0 can be interpreted as receptor-ligand interactions enriched for group combination.Scores
< 0 can be interpreted as receptor-ligand interactions depleted for group combination.[ ]:
rl.interaction_violin_plot(adata,
min_perc=violin_min_perc,
save=f"rl_{violin_out_suffix}.pdf",
figsize=violin_figsize,
dpi=dpi)
7.2 - Network
Show a network graph of number of interactions between groups.
[ ]:
rl.hairball(adata,
min_perc=net_min_perc,
interaction_score=net_interaction_score,
interaction_perc=net_interaction_perc,
save=f"rl_{net_out_suffix}.pdf",
title=net_title,
color_min=net_color_min,
color_max=net_color_max,
restrict_to=net_restrict_to,
show_count=net_show_count
)
7.3 - Receptor-ligand connections
Show a detailed view on receptor-ligand pairs and their strength between groups.
receptor-/ ligand score: Gene enriched (
> 0) or depleted (< 0) for specific group.receptor-/ ligand percent: Percent of cells in a group expressing gene.
interaction score: Receptor-ligand pair enriched (
> 0) or depleted (< 0) between groups.
[ ]:
rl.connectionPlot(adata=adata,
restrict_to=con_restrict_to,
figsize=con_figsize,
dpi=dpi,
connection_alpha="interaction_score",
save=f"rl_{con_out_suffix}.pdf",
title=con_title,
receptor_cluster_col="receptor_cluster",
receptor_col="receptor_gene",
receptor_hue="receptor_score",
receptor_size="receptor_percent",
receptor_genes=con_rec_whitelist,
ligand_cluster_col="ligand_cluster",
ligand_col="ligand_gene",
ligand_hue="ligand_score",
ligand_size="ligand_percent",
ligand_genes=con_lig_whitelist,
filter=con_filter,
lw_multiplier=con_lw_multiplier,
wspace=con_wspace,
line_colors="rainbow")
8 - Get interaction table
[ ]:
interaction_table = rl.get_interactions(adata,
min_perc = tab_min_perc,
interaction_score = tab_interaction_score,
interaction_perc = tab_interaction_perc,
group_a = group_a,
group_b = group_b,
save = f"rl_{tab_out_suffix}.tsv")