Case study: Il1rn knockout bone marrow

This tutorial reproduces the results of IntraTalker – Modelling Intracellular Signalling in Cellular Crosstalk on the murine bone marrow scRNA-seq data of Villatoro et al. (Il1rn KO vs. WT).

The tutorial walks the full pipeline: ligand–receptor inference, transcription factor activities, the receptome, communication communities, and in silico receptor knockout simulations that recover Il1r1 as a driver of the Adipo-CAR → osteolineage transition in mesenchymal stromal cells.

This pipeline uses both Python and R

IntraTalkerPy covers the transcription factor and perturbation modules. The IntraTalker/CrossTalkeR integration runs in R (LR2TF + CrossTalkeR).

The two sides never share objects in memory — they exchange CSV / RDS files on disk. Each crossing is documented in a Handoff box listing exactly which files pass, so you can run either half on its own as long as those files exist.

Workflow Overview

#	Step	Language	Produces
1	Introduction and setup	Python + R	environment, input data
2	Ligand–receptor inference	Python	{KO,WT}_lr_ready.csv
3	Regulon + TF activity	Python	mouse_dorothea_AB.csv, TF activity scores
4	IntraTalker + CrossTalkeR2	R	result_TF_object.RDS, LR_data_final.Rds, edge table
5	Sankey figures	R	Plotting
6	Communication communities	Python	Plotting
7	Receptome construction	Python	MSCs_receptome.csv
8	The MSC trajectory object	Python	MSCs_diffmap_{WT,KO}_subset.h5ad
9	Ridge models + KO simulations	Python	coefficient matrices, delta embeddings
10	Differential pseudotime + ranking	Python	receptor_scores.csv, Plotting

After the setup, the analysis falls into two halves that can be run on their own. Steps 2–6 alternate between Python and R; steps 7–10 are all Python. Each language switch is called out on the page where it happens. The package versions every step was run with are listed under Session info.

Inter- and Intracellular Communication

Steps 2–6. Infer the ligand–receptor interactions and the transcription factor activities, combine both into the intra- and intercellular network with IntraTalker and CrossTalkeR2, and read that network out as Sankey figures and communication communities.

This half is self-contained: if the cell–cell communication analysis is all you are after, you can stop after step 6.

Gene Expression Perturbations

Steps 7–10. Build the receptome, learn receptor to target gene models on the MSC trajectory, knock the receptors out in silico, and rank them by how far they move the cells along pseudotime.

Everything here runs in Python, but it starts from the network of the first half: the receptome is built from the intracellular networks and the differential ligand–receptor table exported from step 4. If you already have MSCs_receptome.csv and the MSC trajectory object from Zenodo, you can jump straight to step 8.

Citation

If you use IntraTalkerPy, please cite:

Klöker V, Nagai JS, Feng Z, Mavrommatis L, Hermanns L, Moscoso JMJ, Ruiz M, Kuppe C, Costa IG. IntraTalker – Modelling Intracellular Signalling in Cellular Crosstalk. bioRxiv (2026). doi: 10.64898/2026.07.17.739174

The data reanalysed in this case study is from Villatoro et al. 2023.