AF: SHF: Small: Adaptive molecular computation using buffered strand displacement networks

学习和适应是生命系统中的基本过程。人类显然能够学习复杂的概念，但即使是细菌等单细胞生物也能够对外部刺激做出适应性反应。这表明，原则上可以开发生物分子的合成系统，使其能够持续运行并表现出类似的适应行为。由于生物技术和纳米技术领域开发的工具的进步，现在有可能在实验室中设计这样的系统。这个项目将研究基于DNA的合成信息处理系统的设计和实验实现，这些系统表现出学习等适应性行为。这项研究的成功完成将使未来能够持续运行的合成生化系统的开发能够具有实际应用，如长期健康监测，以及对生物系统学习的生物化学基础的洞察。该项目将支持本科生和研究生的培训和教育。该项目将研究使用DNA链置换反应实现的合成生化系统中的学习和适应。在DNA链置换反应网络中，多链DNA“门”结构被用来处理使用单链DNA编码的信息，通过与自由扩散的单链结合并作为响应将其他单链释放到溶液中。在这个项目中，将开发一种特殊类型的DNA链位移门，称为“缓冲门”。缓冲门是这样一种门，其中大量门开始处于非活动状态，并且一个子集被激活以执行它们的计算任务：一旦每个门被消耗，除了其通常的输出之外，门还释放一串来激活缓冲器中的新门，因此活动门的数量大致保持不变。在本项目中，缓冲闸门系统将在模拟和实验室中进行研究。至关重要的是，将研究其中一个门控制另一个缓冲门的激活链的可用性的缓冲门系统。这将使信息能够与分子系统一起存储和更新，从而使适应行为成为可能。这一概念将被用作基本的设计构建块，以实现能够响应外部提供的输入信号而自适应行为的分子计算系统。这个项目的目标将是设计、模拟和实现展示自适应行为的分子逻辑系统，例如，实现监督学习算法。

Learning and adaptation are fundamental processes in living systems. Human beings are clearly capable of learning complex concepts, but even unicellular organisms such as bacteria are capable of adaptive response to external stimuli. This suggests that synthetic systems of biomolecules could, in principle, be developed to enable sustained operation and exhibit similar adaptive behaviors. Thanks to advances in the tools developed in the fields of biotechnology and nanotechnology, it is now becoming possible to engineer such systems in the laboratory. This project will study the design and experimental implementation of synthetic DNA-based information processing systems that exhibit adaptive behaviors such as learning. Successful completion of this research will enable future development of synthetic biochemical systems capable of sustained operation, which could have practical applications such as long-term health monitoring, as well as providing insights into the biochemical basis of learning in biological systems. The project will support training and education of undergraduate and graduate students.This project will study learning and adaptation in synthetic biochemical systems implemented using DNA strand displacement reactions. In DNA strand displacement reaction networks, multi-strand DNA "gate" structures are used to process information that is encoded using single strands of DNA, by binding to freely-diffusing single strands and releasing other single strands into solution in response. In this project, a particular class of DNA strand displacement gates known as "buffered gates" will be developed. A buffered gate is one where a large quantity of gates start in an inactive state and a subset are activated to carry out their computational task: once each gate is consumed, in addition to its usual outputs the gate also releases a strand to activate a new gate from the buffer, so the population of active gates remains roughly constant. In this project, buffered gate systems will be studied in simulation and in the laboratory. Crucially, buffered gate systems in which one gate controls the availability of activating strands for another buffered gate will be studied. This will enable information to be stored and updated with the molecular system, thereby enabling adaptive behavior. This concept will be used as the fundamental design building block to implement molecular computing systems capable of adaptive behavior in response to externally-provided input signals. The goal of this project will be to design, simulate and implement molecular logic systems that demonstrate adaptive behavior, for example, implementing supervised learning algorithms.