CytokineLink

Cytokines switch each other on and off. CytokineLink maps those connections as networks, to see how the immune system rewires itself in disease, and why some therapies fail.

Build your own networks

Build cytokine networks from your own data

The CytokineLink pipeline is open source. Give it a single-cell RNA-seq dataset and it returns a validated cytokine–cytokine network for each condition, with the cell types that send and receive each signal. It’s the method behind our ulcerative colitis and Crohn’s disease networks.

  1. Your single-cell data

    A Seurat object with annotated cell types.

  2. Predict interactions

    CytokineLink links cytokines across every pair of cell types.

  3. Healthy
    Inflamed
    Treated
    Non-inflamed

    Your cytokine networks

    One network for each condition, whether healthy tissue, disease states or treatment groups, with the sending and receiving cell types for every interaction. Compare them to see how cytokine communication is rewired.

# get the pipeline
git clone https://github.com/korcsmarosgroup/CytokineLink_pipeline
cd CytokineLink_pipeline/scripts

# run it: Seurat .rds, cell type column, output name
Rscript unified_pipeline.R my_data.rds cell_type my_network

Requires R with tidyverse, Seurat and nichenetr, plus the NicheNet prior model files from Zenodo. Full setup and options are in the repository’s documentation.

Get the pipeline on GitHub Read the method

One cytokine leads to another

Cytokines are small proteins that cells use to talk to each other. When a cell receives a cytokine, it often responds by producing a different one, passing the message on to the next cell.

Each of these chain of events is a cytokine interaction. Below, IFNG from a Th1 cell binds its receptor, IFNGR, on an epithelial cell. Inside the cell, JAK1 and JAK2 activate the transcription factor STAT1, which moves into the nucleus and switches on the CXCL10 gene. The CXCL10 released then attracts more T cells through their CXCR3 receptor.

CytokineLink chains ligand–receptor, signalling and regulatory interactions into a single edge, IFNG → CXCL10, and assembles these edges across cell types into networks. These intracellular cascades are also accessible, and can be explored in detail.

Following a drug through the network

A drug that blocks one cytokine also changes every cytokine that depends on it. In ulcerative colitis, TL1A sits at the top of a cascade: it reaches TNF, IL-1B, IL-23A, CCL20 and IFNG, some directly and some through CSF2, IL-10 and IL-17A.

Watch the signals flow, then block TL1A to see which cytokines it affects. Based on phase 2 trial data from Hassan-Zahraee et al., Inflammatory Bowel Diseases, 2022 Switch to the healthy network to see which routes exist only in UC.

Anti-TL1A target Affected cytokine UC only

Adapted from Olbei et al., Science Signaling, 2026.

Networks and papers

Every network is openly available to explore on NDEx.

  1. 2026 · preprint

    Anti-TNF non-response in Crohn’s disease

    Networks built from single-cell data of Crohn’s disease patients before and after anti-TNF therapy. Non-responders carry an epithelial IL17C-driven inflammatory module before treatment and lose IL10 feedback loops afterwards, sustaining TNF-independent inflammation.

    bioRxiv preprint Code Data
  2. 2026 · DNA and Cell Biology

    From Single Cytokines to Networks: A Systems Immunology View of Ulcerative Colitis

    A review on the future of cytokine networks, and their applications to better understand Ulcerative Colitis.

    Link to paper
  3. 2026 · Science Signaling

    Cytokine networks in ulcerative colitis

    Networks from colonic biopsies of treatment-naive and treatment-exposed ulcerative colitis patients. They recover known interactions, confirmed in patient-derived organoids, and reveal a subnetwork unique to treatment-naive patients.

    Paper Networks: Healthy Inflamed, treatment-naive Inflamed, treatment-exposed Non-inflamed, treatment-exposed
  4. 2024

    Interactions from the Immune Dictionary

    Cytokine–cytokine interactions collated from the supplementary data of Cui et al., 2023.

    Network
  5. 2023

    Interactions from CytoSig

    A signed, directed network of 544 interactions among 155 cytokines, derived from over 2,000 transcriptomics datasets and annotated with connecting intracellular pathways.

    Network
  6. 2021 · Cells

    CytokineLink: the original communication map

    Prevalent cytokines and their receptors assigned to tissues and blood cell types, with case studies in inflammatory bowel disease and COVID-19.

    Paper Network