Domain-Specific Programming Languages for Computational Nucleic Acid Systems

Design and Simulation of a Multilayer Chemical Neural Network That Learns via Backpropagation

The design and implementation of adaptive chemical reaction networks, capable of adjusting their behavior over time in response to experience, is a key goal for the fields of molecular computing and DNA nanotechnology. Mainstream machine learning research offers powerful tools for implementing learning behavior that could one day be realized in a wet chemistry system. Here we develop an abstract chemical reaction network model that implements the backpropagation learning algorithm for a feedforward neural network whose nodes employ the nonlinear "leaky rectified linear unit" transfer function. Our network directly implements the mathematics behind this well-studied learning algorithm, and we demonstrate its capabilities by training the system to learn a linearly inseparable decision surface, specifically, the XOR logic function. We show that this simulation quantitatively follows the definition of the underlying algorithm. To implement this system, we also report ProBioSim, a simulator that enables arbitrary training protocols for simulated chemical reaction networks to be straightforwardly defined using constructs from the host programming language. This work thus provides new insight into the capabilities of learning chemical reaction networks and also develops new computational tools to simulate their behavior, which could be applied in the design and implementations of adaptive artificial life.

Artificial intelligence driven design of catalysts and materials for ring opening polymerization using a domain-specific language

Advances in machine learning (ML) and automated experimentation are poised to vastly accelerate research in polymer science. Data representation is a critical aspect for enabling ML integration in research workflows, yet many data models impose significant rigidity making it difficult to accommodate a broad array of experiment and data types found in polymer science. This inflexibility presents a significant barrier for researchers to leverage their historical data in ML development. Here we show that a domain specific language, termed Chemical Markdown Language (CMDL), provides flexible, extensible, and consistent representation of disparate experiment types and polymer structures. CMDL enables seamless use of historical experimental data to fine-tune regression transformer (RT) models for generative molecular design tasks. We demonstrate the utility of this approach through the generation and the experimental validation of catalysts and polymers in the context of ring-opening polymerization-although we provide examples of how CMDL can be more broadly applied to other polymer classes. Critically, we show how the CMDL tuned model preserves key functional groups within the polymer structure, allowing for experimental validation. These results reveal the versatility of CMDL and how it facilitates translation of historical data into meaningful predictive and generative models to produce experimentally actionable output.

Chemical Reaction Networks’ Programming for Solving Equations

The computational ability of the chemical reaction networks (CRNs) using DNA as the substrate has been verified previously. To solve more complex computational problems and perform the computational steps as expected, the practical design of the basic modules of calculation and the steps in the reactions have become the basic requirements for biomolecular computing. This paper presents a method for solving nonlinear equations in the CRNs with DNA as the substrate. We used the basic calculation module of the CRNs with a gateless structure to design discrete and analog algorithms and realized the nonlinear equations that could not be solved in the previous work, such as exponential, logarithmic, and simple triangle equations. The solution of the equation uses the transformation method, Taylor expansion, and Newton iteration method, and the simulation verified this through examples. We used and improved the basic calculation module of the CRN++ programming language, optimized the error in the basic module, and analyzed the error’s variation over time.

Amphiphilic DNA nanostructures for bottom-up synthetic biology

DNA nanotechnology enables the construction of sophisticated biomimetic nanomachines that are increasingly central to the growing efforts of creating complex cell-like entities from the bottom-up. DNA nanostructures have been proposed as both structural and functional elements of these artificial cells, and in many instances are decorated with hydrophobic moieties to enable interfacing with synthetic lipid bilayers or regulating bulk self-organisation. In this feature article we review recent efforts to design biomimetic membrane-anchored DNA nanostructures capable of imparting complex functionalities to cell-like objects, such as regulated adhesion, tissue formation, communication and transport. We then discuss the ability of hydrophobic modifications to enable the self-assembly of DNA-based nanostructured frameworks with prescribed morphology and functionality, and explore the relevance of these novel materials for artificial cell science and beyond. Finally, we comment on the yet mostly unexpressed potential of amphiphilic DNA-nanotechnology as a complete toolbox for bottom-up synthetic biology - a figurative and literal scaffold upon which the next generation of synthetic cells could be built.

Molecular communication relays for dynamic cross-regulation of self-sorting fibrillar self-assemblies

Structures in living systems cross-regulate via exchange of molecular information to assemble or disassemble on demand and in a coordinated, signal-triggered fashion. DNA strand displacement (DSD) reaction networks allow rational design of signaling and feedback loops, but combining DSD with structural nanotechnology to achieve self-reconfiguring hierarchical system states is still in its infancy. We introduce modular DSD networks with increasing amounts of regulatory functions, such as negative feedback, signal amplification, and signal thresholding, to cross-regulate the transient polymerization/depolymerization of two self-sorting DNA origami nanofibrils and nanotubes. This is achieved by concatenation of the DSD network with molecular information relays embedded on the origami tips. The two origamis exchange information and display programmable transient states observable by TEM and fluorescence spectroscopy. The programmability on the DSD and the origami level is a viable starting point toward more complex lifelike behavior of colloidal multicomponent systems featuring advanced signal processing functions.

Evaluating the effect of ionic strength on PNA:DNA duplex formation kinetics

Peptide nucleic acid (PNA) is a unique synthetic nucleic acid analog that has been adopted for use in many biological applications. These applications rely upon the robust Franklin-Watson-Crick base pairing provided by PNA, particularly at lower ionic strengths. However, our understanding of the relationship between the kinetics of PNA:DNA hybridization and ionic strength is incomplete. Here we measured the kinetics of association and dissociation of PNA with DNA across a range of ionic strengths and temperatures at single-molecule resolution using total internal reflection fluorescence imaging. Unlike DNA:DNA duplexes, PNA:DNA duplexes are more stable at lower ionic strength, and we demonstrate that this is due to a higher association rate. While the dissociation rate of PNA:DNA duplexes is largely insensitive to ionic strength, it is significantly lower than that of DNA:DNA duplexes having the same number and sequence of base pairing interactions. The temperature dependence of PNA:DNA kinetic rate constants indicate a significant enthalpy barrier to duplex dissociation, and to a lesser extent, duplex formation. This investigation into the kinetics of PNA:DNA hybridization provides a framework towards better understanding and design of PNA sequences for future applications.