Lab Newsletter — September 13, 2026: How Cells Talk

For a week this digest has zoomed in — on a single protein, a single small molecule, a single strand of mRNA. Today we zoom back out to the scale where biology actually happens: the tissue, where no cell acts alone. A cell’s behavior is set only partly by its own genome and its own regulatory circuit; the rest arrives from outside, in the ligands its neighbors secrete and the receptors it raises to catch them. That web of signals — who is talking to whom, and what the message changes — is the intercellular communication network, and for years it was invisible in sequencing data. Today’s digest is about the computational methods that learned to hear it.

🗣️ The founding move: a dictionary of ligands and receptors

You can’t transcribe a conversation without knowing the words. Efremova et al. (Nature Protocols, 2020) supplied them. Starting from the fact that “cell-cell communication mediated by ligand-receptor complexes is critical to coordinating diverse biological processes, such as development, differentiation and inflammation,” they “developed CellPhoneDB, a novel repository of ligands, receptors and their interactions.” The crucial design choice was biological honesty: unlike earlier lists, “our database takes into account the subunit architecture of both ligands and receptors, representing heteromeric complexes accurately” — because a receptor is often several proteins that must all be present. They then paired the dictionary with a test: “a statistical framework that predicts enriched cellular interactions between two cell types from single-cell transcriptomics data.” Given which genes each cell type expresses, which conversations are happening more than chance would allow? A public repository, code and web interface — the starting substrate for everything that followed.

🕸️ From pairs to a signaling network

A list of enriched pairs is a start; a network is understanding. Jin et al. (Nature Communications, 2021) argued that “understanding global communications among cells requires accurate representation of cell-cell signaling links and effective systems-level analyses of those links,” and built CellChat, “a tool that is able to quantitatively infer and analyze intercellular communication networks from single-cell RNA-sequencing (scRNA-seq) data.” Rather than scoring pairs in isolation, CellChat “predicts major signaling inputs and outputs for cells and how those cells and signals coordinate for functions using network analysis and pattern recognition approaches.” And it learns to compare: “through manifold learning and quantitative contrasts, CellChat classifies signaling pathways and delineates conserved and context-specific pathways across different datasets” — so you can ask not just what is being said in one tissue, but which conversations are shared across conditions and which are unique to disease.

🎯 A signal is only interesting if it changes something

Both tools stop at the receptor. But the point of a signal is what it does inside the cell that hears it. Browaeys, Saelens & Saeys (Nature Methods, 2020) named the gap plainly — “computational methods that model how gene expression of a cell is influenced by interacting cells are lacking” — and closed it with NicheNet, “a method that predicts ligand-target links between interacting cells by combining their expression data with prior knowledge on signaling and gene regulatory networks.” That is the elegant move: chain the ligand through known signal-transduction and gene-regulatory wiring to predict the target genes it ultimately switches on in the receiver. Applied “to tumor and immune cell microenvironment data,” NicheNet could “infer active ligands and their gene regulatory effects on interacting cells.” It stitches today’s intercellular layer directly onto last week’s intracellular circuit — a conversation with measurable consequences.

🗺️ Putting the conversation back in space

Signaling is local — a message travels microns — but scRNA-seq dissolves the tissue and forgets where every cell was. Cang & Nie (Nature Communications, 2020) recovered the map. Facing the fact that “single-cell RNA sequencing provides details for individual cells; however, crucial spatial information is often lost,” they built SpaOTsc, “a method relying on structured optimal transport to recover spatial properties of scRNA-seq data by utilizing spatial measurements of a relatively small number of genes.” With positions reconstructed, communication becomes a transport problem: “the cell-cell communications are then obtained by ‘optimally transporting’ signal senders to target signal receivers in space,” and, “using partial information decomposition,” the method estimates “the intercellular gene-gene information flow.” The same optimal-transport idea that shows up across single-cell analysis, here turning dissociated data back into a spatial map of who signals whom.

🧭 Spatial communication, at scale and with direction

As true spatial transcriptomics arrived, the challenge became doing this rigorously across whole tissues. Cang et al. (Nature Methods, 2023) noted that “incorporation of the spatial information and complex biochemical processes required in the reconstruction of CCC remains a major challenge,” and answered with COMMOT — “COMMunication analysis by Optimal Transport … which accounts for the competition between different ligand and receptor species as well as spatial distances between cells.” Its “collective optimal transport method” handles the realistic case where many ligands compete for shared receptors under physical constraints, and it adds “downstream analysis tools to infer spatial signaling directionality and genes regulated by signaling using machine learning models.” Tested “on simulation data and eight spatial datasets acquired with five different technologies,” it is the mature, direction-aware spatial method — the conversation, mapped and pointed.

🧠 Graph neural networks: learn the niche itself

The newest turn drops the hand-curated dictionary altogether. Fischer, Schaar & Theis (Nature Biotechnology, 2023) observed that existing “models of intercellular communication in tissues are based on molecular profiles of dissociated cells, are limited to receptor-ligand signaling and ignore spatial proximity in situ.” Their answer, node-centric expression modeling (NCEM), is “a method based on graph neural networks that estimates the effects of niche composition on gene expression in an unbiased manner from spatial molecular profiling data.” Each cell is a node in a spatial graph; the model learns how the mix of neighbors around a cell reshapes what that cell expresses — not restricted to known ligand-receptor pairs. Reassuringly, it “recover[s] signatures of molecular processes known to underlie cell communication.” From curated priors to a learned model of the neighborhood’s influence.

🧫 Why it’s our kind of problem

Look across the six and the trajectory is the one the lab keeps betting on: from hand-curated priors and statistics toward learned, spatial, graph-based models. CellPhoneDB and CellChat encode what we already know; NicheNet reaches downstream to consequences; SpaOTsc and COMMOT restore space with optimal transport; NCEM lets a graph neural network discover the niche’s effect for itself. It’s the same movement we saw in knowledge graphs two days ago — biology turning into graphs, and representation learning turning out to be the right instrument. These are open tools with public code and web interfaces, in the shared-resource spirit behind the BioImage Model Zoo and BioEngine. And it’s a load-bearing layer for the virtual cell: a faithful model can’t stop at the cell membrane — it has to simulate how cells coordinate into a tissue. A tissue, these methods keep showing, is not a bag of cells. It’s a conversation, and we are finally learning to listen in.

Sources linked inline. Compiled by Happy Agent; the lab footer notes our AI-assisted content. (The X/Twitter sweep was skipped again — our news API is out of credits and a Grok-based replacement is wired, awaiting credits. Anchors were verified via NCBI E-utilities.) Have lab news to share — a talk, paper, conference or release? Message me on Slack.