Noise Control for DNA Computing

Tomislav Plesa, Konstantinos C. Zygalakis, David F. Anderson, Radek ErbanarXiv 2017 · DOI · arXiv

Abstract

Synthetic biology is a growing interdisciplinary field that aims to design biochemical systems with specific, desired behaviors. With the advancement of strand-displacement DNA computing, a large class of abstract biochemical networks can now be physically realized using DNA molecules. However, systematic design methods have predominantly focused on the deterministic level, neglecting the stochastic effects that play a crucial role in biochemistry at small scales. In such environments, controlling intrinsic noise is necessary for successful network design. This paper introduces a noise-control algorithm that structurally modifies reaction networks under mass-action kinetics. The algorithm introduces controllable state-dependent noise into the stochastic dynamics while preserving the original deterministic dynamics. The capabilities of this approach are demonstrated on a production-decay system, where it achieves noise-induced multistability, and an exotic system, where it controls stochastic switching to achieve noise-induced oscillations.

Introduction

In the field of synthetic biology, the goal is to construct molecular "circuits" that behave predictably. Traditionally, these circuits are designed using deterministic models—mathematical frameworks that describe the average concentration of molecules over time. However, at the molecular scale of a cell or a test tube, molecules do not move in smooth averages; they collide and react randomly. This "intrinsic noise" becomes a dominant force when molecule counts are low, often leading to behaviors that deviate significantly from deterministic predictions.

While noise is frequently viewed as a nuisance to be suppressed, biological systems often utilize it functionally—for example, to facilitate switching between different cellular states. The primary challenge in engineering these systems is that in standard chemical reaction networks (CRNs), the noise and the average behavior are "coupled." Changing a reaction to adjust the noise typically alters the average behavior as well. This research introduces a systematic method to decouple these two factors, allowing for independent control over a system's mean and its variance.

The Noise-Control Algorithm

The researchers propose a three-step algorithm to modify any given chemical reaction network to achieve specific stochastic goals without altering the underlying deterministic path:

  1. Pairwise Conservation: For every species $s$ in the original network, a "shadow" species $\bar{s}$ is introduced such that the total count ($s + \bar{s}$) remains constant. This creates a bounded environment for the dynamics to unfold.
  2. Drift Correction: Auxiliary species and fast reactions (a "drift-corrector network") are added to ensure that the new, expanded network follows the exact same deterministic trajectory as the original system in the limit of specific parameters.
  3. Zero-Drift Networks: This component allows for the injection of state-dependent noise. By embedding specific sets of reactions that fire in opposite directions with equal rates, the researchers can add "stochastic kicks" to the system. Deterministically, these reactions cancel out (zero drift), but stochastically, they increase variance. By choosing where in the state-space these reactions are most active, the probability distribution of the system can be shaped to favor specific states.

Crucially, these modifications are designed to be physically realizable. While the theoretical modifications may involve complex multi-molecular collisions, they can be decomposed into simple two-molecule reactions, making them compatible with DNA strand-displacement techniques.

Experimental Benchmarks

As a theoretical and computational study, the algorithm was evaluated using two "in silico" benchmark systems to test its versatility:

1. Production-Decay System

In a simple one-species model, the algorithm was used to demonstrate how noise can create "multistability." The original network naturally possessed only one stable state. By applying the noise-control algorithm, the researchers redesigned the network to be "bimodal" (flipping between two states) and even "trimodal" (flipping between three), all while maintaining the same average concentration of molecules.

2. Exotic Two-Species System

The algorithm was also tested on a complex 11-reaction network known to exhibit bistability (a stable point versus a limit cycle). In the original stochastic model, the system spent most of its time near a single equilibrium point, rarely entering the oscillatory "limit cycle." The noise-control algorithm allowed researchers to "rebalance" the system, forcing it to spend more time in the oscillatory state or inducing oscillations where they were previously rare.

Discussion and Insights

The core mathematical insight of this work is the power of "Zero-Drift" modules. By treating noise as a programmable feature rather than a bug, the paper demonstrates that researchers can create "noise-induced" phenomena that have no deterministic counterpart. This provides a formal bridge between the relatively straightforward world of deterministic design and the complex, stochastic reality of molecular computing.

However, there are practical considerations for physical implementation. The "noise controllers" require the addition of several new species and reactions, which increases the potential for "leakage" and side reactions in a physical DNA computer. Furthermore, the algorithm relies on certain reactions being significantly faster than others (the $\mu \to 0$ limit). In real-world DNA chemistry, physical speed limits may affect how perfectly the deterministic behavior is preserved.

Conclusion

The noise-control algorithm represents a significant step forward for DNA-based synthetic biology. It allows engineers to first define a desired average behavior and then layer on noise-control modules to achieve specific stochastic functions, such as probabilistic logic gates or molecular timers. By decoupling the mean from the variance, this method transforms molecular noise into a powerful tool for computational design.

Summary updated

Abstract

Synthetic biology is a growing interdisciplinary field, with far-reaching applications, which aims to design biochemical systems that behave in a desired manner. With the advancement of strand-displacement DNA computing, a large class of abstract biochemical networks may be physically realized using DNA molecules. Methods for systematic design of the abstract systems with prescribed behaviors have been predominantly developed at the (less-detailed) deterministic level. However, stochastic effects, neglected at the deterministic level, are increasingly found to play an important role in biochemistry. In such circumstances, methods for controlling the intrinsic noise in the system are necessary for a successful network design at the (more-detailed) stochastic level. To bridge the gap, the noise-control algorithm for designing biochemical networks is developed in this paper. The algorithm structurally modifies any given reaction network under mass-action kinetics, in such a way that (i) controllable state-dependent noise is introduced into the stochastic dynamics, while (ii) the deterministic dynamics are preserved. The capabilities of the algorithm are demonstrated on a production-decay reaction system, and on an exotic system displaying bistability. For the production-decay system, it is shown that the algorithm may be used to redesign the network to achieve noise-induced multistability. For the exotic system, the algorithm is used to redesign the network to control the stochastic switching, and achieve noise-induced oscillations.
BibTeX
@misc{plesa2017noise,
  author = {Tomislav Plesa and Konstantinos C. Zygalakis and David F. Anderson and Radek Erban},
  title = {{Noise Control for DNA Computing}},
  year = {2017},
  eprint = {1705.09392},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1705.09392},
}
Figure