| With the continuous reduction in chip size,semiconductor technology is gradually approaching its physical limits.Electronic computers may not be able to meet the computing power requirements of future technological developments.Therefore,exploring new computational methods and finding new engineering materials is imperative to break through this bottleneck.DNA molecules have advantages such as easy availability of materials,strong storage capacity,low energy consumption,and fast parallel computing speed,making them an ideal engineering material for biological computing.DNA circuits are an important component of biological computers,and in recent years,DNA circuits have become a hot research topic.With the development of science and technology,more and more engineering fields have put forward higher requirements for the functions of DNA circuits,the tasks to be performed are becoming increasingly complex,and the scale of DNA circuits is also becoming larger and larger.Therefore,optimizing the performance of large-scale DNA circuits to enable them to perform more complex and diverse tasks is crucial.This article conducts in-depth research around critical issues such as the scale and complexity of DNA circuits and the expansion of application fields to optimize traditional DNA circuit models and expand the circuit scale while simplifying the complexity of DNA circuits,thereby improving the response speed of circuits.In addition,combining DNA circuits with artificial neural networks expands the application of circuits in molecular pattern recognition.The main contents are as follows:Aiming at the problems of complex circuits,a large number of DNA strands,and long reaction time when implementing circuits using traditional dual-rail logic methods,molecular switching strategies are used to build large-scale DNA circuits to improve the above problems.Firstly,two small-scale digital logic circuits were constructed,and a DNA switching circuit with a two-bit parity device in one of the small-scale circuits was compared with a dual-rail logic circuit.The simulation results showed that the reaction time required for dual-rail was tens of times that of a DNA switching circuit,and the number of DNA strands required was 2.5times that of a switching circuit.In order to further demonstrate the advantages of DNA molecular switches in implementing large-scale digital circuits,this paper constructs a 16-line-4-line priority encoder for large-scale digital logic circuits using DNA molecular switches.In order to enable DNA circuits to be applied to molecular recognition,attempts have been made to use DNA circuits to build artificial neural networks.A two-competitors neural network model capable of realizing recognition functions was established by combining the winner take all strategy and DNA strand replacement technology.The model was used to identify target patterns of 4 and 144 bits,respectively.In recognizing 144 bits target patterns,extracting feature values and grouping simplifies the recognition process,reduces the use of DNA strands,and proves the ability of two competitive neural networks to recognize input information.By designing a particular competition template chain that allows three competitors to participate in competition simultaneously,the winner takes all neural network model has been further optimized.Using this competition template chain in a neural network containing a winner takes all strategy,a three-competitors neural network model was successfully established.This model can process and recognize more patterns at approximately the same time as a two-competitors neural network.The successful establishment of this model is expected to apply DNA neural networks to more fields and complete the processing of more information.In this paper,the DNA circuit model has been optimized based on DNA strand displacement technology,constructed large-scale DNA molecular switching circuits,combined with the winner takes all strategy to build a two-competitors winner takes all neural network,and designed a new competitive template chain to optimize the neural network model further.It not only provides a new method for DNA circuits to implement large-scale digital logic circuits but also provides a reference for DNA circuits to achieve more functions and be used in more fields. |
