Cell Growth & Morphogenesis: How a Cluster Builds Its Own Shape

Growth as a physical packing problem

Strip away the biochemistry and a growing cell cluster is, at its core, a mechanics problem: soft, roughly circular objects that enlarge over time, occasionally split into two, and constantly jostle their neighbours for space. This simulation models each cell as a disc in a force-directed system — every overlapping pair of cells pushes apart with a force that grows the more they overlap, while cells close enough to touch also feel a mild attraction that keeps the cluster cohesive rather than flying apart. Let a few seed cells grow and divide under these rules and the cluster self-organises into a tightly packed, roughly hexagonal arrangement, the same close-packing geometry that soap bubbles, ball bearings and, not coincidentally, real epithelial tissue all settle into when many similarly-sized soft objects are squeezed together.

Division rate and adhesion set the shape

Two parameters dominate how the cluster's shape evolves. The division rate controls how quickly cells reach a size threshold and split, and a faster rate produces a cluster that grows and roughens more quickly, since new divisions constantly disturb the local packing before it can relax. Adhesion, the strength of the mild attractive force between nearby cells, controls how tightly the cluster holds together — high adhesion keeps the boundary smooth and roughly circular as new cells are absorbed into a compact mass, while low adhesion lets the cluster's edge become ragged and finger-like, since newly divided cells can drift away from their neighbours before the repulsive packing forces pull the bulk back into shape. Both parameters interact with the packing forces described above; changing either one changes not just the growth rate but the entire qualitative shape the cluster settles into.

Turing's other idea: chemistry that paints patterns

Growth explains the cluster's shape, but not its internal structure — real tissue is not uniform, it has patterned regions of different cell types. Alan Turing's 1952 paper on morphogenesis proposed a strikingly simple chemical mechanism for exactly this: two diffusing substances, an activator that promotes its own production and a faster-diffusing inhibitor that suppresses the activator, can spontaneously organise a uniform field into stable spots, stripes or other regular patterns purely from random fluctuations, with no external template needed. The Gray-Scott model is one of the best-known concrete implementations of this idea, describing two chemicals U and V reacting and diffusing across a grid:

∂U/∂t = Du ∇²U − U V² + f (1 − U)
∂V/∂t = Dv ∇²V + U V² − (f + k) V

Du, Dv  — diffusion rates (Du > Dv, the activator spreads slower)
f       — feed rate: how fast U is replenished
k       — kill rate: how fast V is removed

Depending on the feed and kill rates, the same two equations produce dramatically different steady patterns — spots, stripes, maze-like labyrinths or slowly pulsing blobs — from a system that starts as almost featureless random noise. The pattern is not designed anywhere; it emerges purely from the interplay of local reaction and differential diffusion, which is exactly why Turing's mechanism is considered a plausible physical explanation for pattern formation in real biology, from the spots on a leopard to the ridges on a fish's fin.

Coupling the pattern to the growing cluster

In this simulation the reaction-diffusion field runs across the same space the cell cluster occupies, and the local morphogen concentration a cell experiences can bias its behaviour — for instance nudging its division rate or marking a region of cells as a distinct fate. This mirrors a foundational idea in developmental biology called positional information: cells in a real embryo do not know in advance what tissue type they will become, they read the local concentration of diffusing signalling molecules called morphogens, such as Sonic hedgehog or members of the BMP family, and different concentration thresholds trigger different genetic programs. A morphogen gradient turns a spatial position into a piece of information a cell can act on, letting a genetically uniform population of cells differentiate into a structured, patterned tissue.

Two feedback loops, one emergent structure

What makes the combined simulation interesting is that the two processes feed back on each other: the growing, dividing cluster changes the domain the reaction-diffusion pattern lives on, reshaping where activator and inhibitor can spread, while the resulting pattern can in turn bias where and how fast cells divide. Neither the packing mechanics nor the chemistry alone produces the layered, patterned structure that emerges when both run together — a small, concrete illustration of how mechanical and chemical self-organisation combine in real developmental biology to build a structured organism from a single starting cell, with no external blueprint anywhere in the system.

Frequently asked questions

Try it live

Everything above runs in your browser — open Cell Growth & Morphogenesis and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.