We tend to separate the brain and the muscles; the brain thinks; the muscle does its job. The brain perceives complex information about the world and makes decisions, while the muscles just work. This distinction extends to our understanding of cellular processes, where some molecules within the cell are viewed as “thinkers” that process information from the chemical environment to determine the actions necessary for survival, while others are viewed as “muscles” that form basic structures. For the survival of the cell.

However, a new study shows how the molecules that make up the structures, that is, muscles, can both think and move on their own. The research, conducted by scientists from Maynooth University, University of Chicago and Cal Institute of Technology, was published in the journal. Nature.

“We show that a natural molecular process (nucleation) long studied as a ‘muscle’ can perform complex computations that rival a simple neural network,” said Associate Professor Arvind Murugan of the University of Chicago. writers. “This is a clearly hidden ability that evolution can use in cells to do more with less; “Making” molecules can also be “thinking”.

Thinking with the help of physics

In order to survive, cells must recognize their environment and take different actions. For example, some combinations of molecules may indicate times of stress that require squatting, while other combinations of molecules may indicate times of abundance. However, the difference between these molecular signals can be very subtle; Different environments may contain the same molecules in different proportions.

Dr D., a research fellow at Maynooth University’s Hamilton Institute and lead author of the study. Constantine Evans explained that it was like walking into a house and smelling freshly baked biscuits instead of burning tyres. “Your brain will change your behavior depending on how you experience different combinations of odorous chemicals. We decided to ask whether molecular systems physics could do the same thing, even though they don’t have brains,” he said.

The conventional wisdom is that cells can sense and respond in this way using molecular circuits that are conceptually similar to the electronic circuits of your laptop; some molecules feel, other molecules decide what to do, and finally “muscle” molecules perform the action (e.g. building a structure).

An alternative idea considered here is that all of these tasks (sensing, deciding, reacting) can be accomplished in a single step by the physics inherent in the “muscle” itself. This research involves the physics of “phase transitions”; Imagine a glass of water that freezes when it reaches 0 °C; it “starts” as a small piece of ice and then grows until the entire glass of water is frozen.

At first glance, these first steps of the act of “freezing” – birth – do not resemble “thinking”. But this study shows that the action of freezing can “recognize” subtle chemical combinations, like the scent of oatmeal raisin cookies or chocolate chips, and create different molecular structures in response.

Reliability in experiments

The authors tested the robustness of nucleation-based decision-making using DNA nanotechnology, a field pioneered by Professor Eric Winfrey. “The theory is general and should apply to all kinds of molecules. But DNA allows us to experimentally study nucleation in complex mixtures of thousands of types of molecules and systematically understand the impact of how many types of molecules are present and what types of interactions they have,” Eric explained.

The experiment revealed a few surprises; muscle-based decision making was surprisingly reliable and scalable. During the experiment, it became clear that complications not modeled in theory, such as missing molecules, helped rather than harmed. As a result, relatively simple experiments solved pattern recognition problems involving about a thousand types of molecules, almost 10 times better than previous pattern-based approaches. In each case, the molecules came together to form different nanometer-scale structures in response to different chemical patterns; but the act of building the structure determined what was to be built.

The work points to a new way of looking at computing that involves not designing a circuit, but what physicists call a “phase diagram”; for example, a phase diagram for water might describe the temperature and pressure conditions under which liquid water freezes or boils. Traditionally, phase diagrams are thought to describe material properties similar to “muscles”. However, this study shows that the phase diagram can encode “thinking” as well as “doing” when scaling to complex systems with many different properties. species components.

“Physicists have traditionally studied things like a glass of water containing many molecules, but they are all the same. But a living cell is full of many different cells species Molecules that interact with each other in complex ways. This opens up distinct new possibilities for multicomponent systems,” said Dr. Jackson O’Brien, who participated in the research as a graduate student in physics at the University of Chicago. The theory in this work draws mathematical analogies between such multicomponent systems and the theory of neural networks; experiments showed how these multicomponent systems could be made to work Refine computational properties through a physical process, similar to how the brain learns to associate different smells with different actions.

Although the experiments here concern DNA molecules in a test tube, the basic concepts of nucleation in systems with many types of components are broadly applicable to many other molecular and physical systems. The authors hope that this work will stimulate studies aimed at uncovering hidden “thinking” capabilities in other multicomponent systems that currently appear to be merely “muscles.”