Enzyme-Free Scalable DNA Digital Design Techniques: A Review

Abstract

With the recent developments in DNA nanotechnology, DNA has been used as the basic building block for the design of nanostructures, autonomous molecular motors, various devices, and circuits. DNA is considered as a possible candidate for replacing silicon for designing digital circuits in a near future, especially in implantable medical devices, because of its parallelism, computational powers, small size, light weight, and compatibility with bio-signals. The research in DNA digital design is in early stages of development, and electrical and computer engineers are not much attracted towards this field. In this paper, we give a brief review of the existing enzyme-free scalable DNA digital design techniques which are recently developed. With the developments in DNA circuits, it would be possible to design synthetic molecular systems, therapeutic molecular devices, and other molecular scale devices and instruments. The ultimate aim will be to build complex digital designs using DNA strands which may even be placed inside a human body.

Principles and Applications of Nucleic Acid Strand Displacement Reactions

“…More recent work in enzyme-free nucleic acid dynamical systems has been discussed by Srinivas et al, whereas the review by Bi, Yue, and Zhang focuses on hybridization chain reactions and their application to biosensing, bioimaging, and biomedicine. A recent review focusing on the design of enzyme-free DNA reaction networks that carry out computation has been given by George and Singh …”

“…A recent review focusing on the design of enzyme-free DNA reaction networks that carry out computation has been given by George and Singh. 86 We now proceed with a discussion of the physical chemistry of strand displacement reactions in order to describe the elemental reaction mechanisms employed in all-DNA dynamic DNA systems. Since sometimes the purpose of these reactions is to perform mechanical work, the mechanochemistry of DNA hybridization is also discussed.…”

DNA strand displacement‐based logic inverter gate design

“…Among these logic gate designs, all of them are not suitable for the design of large complex circuits. A brief review of all the scalable digital DNA designs is given in [9]. Digital circuits made up of DNA strands can be used in nanomachines and devices such as DNA nanorobots [10][11][12].…”

Constructing artificial neural networks using genetic circuits to realize neuromorphic computing

“…(Very-large-scale integration, VLSI)来模仿人脑中神经网络的计算结构。 [1][2][3] 其 核心思想是以数学模型来模拟人类神经元(Neuron)及突触(Synapse)的结构， 并结合多层次传导来模拟大量神经元联结。 [4,5] 这种仿生学方法创造了高度连接 的合成神经元和突触结构，从而可以执行复杂的计算任务。 [6] 利用人工神经网 络实现神经拟态计算已在深度学习中取得了空前的成功，并广泛应用于人工智 能的各种研究方向。 [7][8][9] 其如何定量、准确、简化地描述神经元和突触，直接决 定了计算系统的性能、功耗与设计复杂度。 与基于硅材料的传统计算系统不同，以 DNA (Deoxyribonucleic Acid) 分子等 生物材料为基础的生物计算，因具有超低能耗并行处理信息的能力而备受关注。 [10][11][12][13] 统如何表现出类脑计算的自主行为，使计算形式得到了革命性的突破。 [14][15][16][17][18][19][20] 使 用简单的 DNA 逻辑门结构，可以系统地将人工神经网络转换为 DNA 置换链级 联反应，并可以执行类似大脑的学习和记忆等操作。 [14] 这展示了利用生物材料 构造人工神经网络的巨大发展前景，并将促进神经拟态计算的广泛部署。然而， 通过 DNA 置换链级联反应构建神经网络也面临诸多挑战。 [21] 首先，DNA 链置 换反应使用为特定计算而设计的预编程一次性体系，执行不同逻辑电路任务将 需要不同的一组 DNA 分子；其次，DNA 或酶反应多为不可逆反应，导致逻辑 门不能反复地"打开"和"关闭"，这从根本上限制了 DNA 计算电路从单层扩 展到多层结构。 [22][23][24][25][26] 基因线路作为一种新型的 DNA 计算方式为解决上述问题提供了新的思路。 在利用基因线路构建的人工神经网络中，基因电路元件称为动态单元，这些动 态单元可以随着输入或输出的不同而反复使用，并可动态地将上游电路的输出 直接连接到下游电路的输入，从而实现更复杂的逻辑模式识别。 [27][28][29] 在本研究 中，我们通过基因线路设计构建了人工神经网络，并实现了线性分类、非线性 测量荧光值并以相对启动子单元(Relative Promoter Unit，RPU)表示 [30] 。本研 究中使用的生物元件库来源于 Genetic circuit design automation [30] (补充材料表 输出的非线性变换。 [31] 以记住两个模式"L"和"T"。网络中存在三个输入，通过不同的排列组合，可 以形成八种不同的"内存"(如图 6(c)所示)。这八种"内存"是含有噪声的 "L"和"T"。网络通过将"内存"与目标模式进行比较，用于判断"内存"与 目标模式中的"L"和"T"中的哪个模式最为相似。判断的依据是"内存"分别 与目标模式进行权重和运算，计算结果越大，则"内存"与该目标模式越相近， 若计算结果相同，则认定"内存"为目标模式"L"。图 6(c)中所包含八种不同 的"内存"中，三种与目标模式"T"情况相似，五种与目标模式"L"情况相似。 断"内存"的分类情况。图 7(c)仿真图纵坐标的 RPU 代表"内存"的分类情况。 在图 7(c)中当输入为"000"、"010"、"100"、"101"、"110"，此时输出元件 Y5 的 RPU 较低，表示输出信号为"0"，即在这五种输入情况下构成的"内存"…”