Connectivity-dependent effects of depth-1 canalisation on attractor dynamics in small-world Boolean networks

Life is built on a vast, intricate web of instructions. Inside every cell, genes act as switches, turning on and off to tell the organism how to grow, respond to its environment, and maintain its health. These switches do not operate in isolation; they form a network where one gene's activity can trigger or silence another. Scientists have long sought to understand how the shape of this network—the way the genes are wired together—determines the stability of the whole system. If the network is too chaotic, the cell might fail; if it is too rigid, it cannot adapt. To study this, researchers often use simplified computer models called Boolean networks. In these models, every gene is reduced to a simple binary state: either on or off. The rules that govern how these genes change state are like the logic of a circuit, where the combined input from neighboring genes decides the next move.

A key feature of real biological systems is that they often have "canalising" rules. This means that sometimes, a single dominant signal can override all other inputs and force a gene into a specific state, regardless of what the rest of the network is doing. It is a bit like a master switch that, when flipped, cuts power to an entire room, ignoring the status of every other light switch in the house. Scientists have known for some time that these dominant rules tend to make networks more stable, often shortening the time it takes for the system to settle into a repeating pattern, known as an attractor. However, a new study by researchers at the University of Queensland and the University of Bisha suggests that this stability is not a universal rule. Instead, the effect of having a dominant switch depends entirely on how the network is connected.

The researchers built millions of virtual networks to test this idea. They created digital systems where nodes, representing genes, were connected in a specific pattern known as a small-world network. This structure mimics real biological systems by having most connections clustered locally, with a few long-range links that bridge distant parts of the network. They then programmed these networks with two types of logic: standard rules where all inputs are weighed equally, and canalising rules where one input can dominate. They ran these simulations across a wide range of conditions, changing the size of the network, the density of connections, and the probability that a connection would be an inhibitor (a signal that turns a gene off) rather than an activator.

The results revealed a surprising twist. In most of the scenarios they tested, adding the dominant, canalising rules did indeed shorten the time it took for the network to settle into a stable loop, confirming the traditional view that these rules promote order. However, in networks with very low connectivity—where each gene was linked to only one or two neighbors—the effect flipped. In these sparse networks, the dominant rules actually made the system take longer to settle, increasing the length of the repeating cycle by roughly twenty to twenty-six percent. This reversal showed that the benefit of having a master switch is not absolute; it depends on the surrounding web of connections. If the network is too sparse, the dominant rule disrupts the flow of information in a way that prolongs the system's journey to stability.

The study also looked at how these networks recover when disturbed. In the real world, cells are constantly bombarded by internal and external shocks. The researchers simulated this by flipping the state of a single gene in their models and watching what happened next. They found that networks with the dominant canalising rules were much better at recovering. When a single gene was flipped, these networks were more likely to return to their original stable pattern within a short time, and the disturbance spread less far through the system. This suggests that while the dominant rules might sometimes lengthen the time to settle in very sparse networks, they generally make the system more robust against errors and better at returning to its intended state.

The researchers also examined how the balance of inhibitory signals affected the outcome. They found that the presence of inhibitory connections was crucial. When the network had a higher probability of inhibitory links, the dominant rules were much more effective at shortening the settling time. However, when inhibitory links were rare, the effect of the dominant rules was weaker and, in the case of low connectivity, reversed. This indicates that the interplay between the type of rule and the type of connection is complex. The dominant rule does not work in a vacuum; its impact is shaped by the overall architecture of the network and the balance of signals flowing through it.

One of the most significant findings was that the change in stability was not simply because the networks were settling into a single, unchanging state. The researchers measured how often the system ended up in a fixed point where nothing changed at all, and they found that the dominant rules actually reduced the frequency of these static states. Instead, the networks were settling into shorter, more efficient repeating cycles. This distinction is important because it shows that the system is not just freezing into place; it is finding a more efficient rhythm. The study also found that increasing the number of random long-range connections in the network generally shortened the settling time, but this effect varied depending on the specific rules and the density of the network.

The implications of this work extend beyond computer models. The researchers suggest that these findings help explain why certain genes, which act as strong local regulators in real organisms, might have different effects in different genetic backgrounds. A gene that stabilizes a trait in one variety of a plant might have a neutral or even destabilizing effect in another, depending on the connectivity of the underlying regulatory network. The study does not claim to have solved the mystery of biological stability, but it provides a clear, simulated demonstration that the context of the network matters. The same rule can lead to order or chaos depending on the connections it sits within.

Ultimately, the paper concludes that we cannot understand the behavior of a regulatory system by looking at the rules in isolation. The stability of a cell, or the resilience of a complex trait, emerges from the combination of the local logic and the global structure. Whether a dominant signal acts as a stabilizer or a disruptor depends on the density of the network, the balance of activating and inhibiting signals, and the presence of long-range connections. For scientists trying to engineer better crops or understand complex diseases, this means that simply identifying a key regulator is not enough. One must also understand the wiring diagram of the system in which that regulator operates, because the effect of a single switch is written by the network it controls.