AI 中文总结
该研究针对工程化DNA系统的平衡浓度预测问题，通过定义帕累托最优聚合物并开发结合单体数量限制与组合覆盖设计的可扩展框架，实现了比直接计算快数量级的平衡相关聚合物枚举。
AI 中文摘要
预测分子复合物的平衡浓度对于验证工程化DNA系统的行为至关重要。然而，有限的单体类型原则上可生成无限多的复合物。我们在一种无几何的结构域水平抽象模型（称为结构域-单体系统）中研究此候选枚举问题，该模型将热力学绑定网络（TBNs）推广至非饱和设置，其中无需形成所有可能的键。我们将帕累托次优聚合物定义为可拆分为非相互作用部分的聚合物，并证明仅关注帕累托最优聚合物具有热力学合理性：任何帕累托次优聚合物都不会出现在任何最小自由能构型中，且此类聚合物的总平衡浓度很小。我们证明帕累托最优聚合物数量有限，并通过希尔伯特基计算对其进行精确表征，扩展了先前针对饱和TBN模型的研究。为将此方法扩展至大型系统，我们开发了一种框架，该框架限制单个聚合物包含的不同单体类型数量，并使用组合覆盖设计减少所需的希尔伯特基计算次数。我们在多个DNA分子编程系统系列上对该方法进行基准测试，结果显示其相较于直接计算实现了数量级的加速，同时恢复了几乎所有与平衡相关的聚合物。
英文摘要
Predicting equilibrium concentrations of molecular complexes is essential for verifying the behavior of engineered DNA systems. However, a finite set of monomer types can in principle generate infinitely many complexes. We study this candidate-enumeration problem in a geometry-free, domain-level abstraction called a domain-monomer system, generalizing Thermodynamic Binding Networks (TBNs) to the unsaturated setting where not every possible bond need be formed. We define Pareto-suboptimal polymers as those that can be split into non-interacting parts, and show that restricting attention to Pareto-optimal polymers is thermodynamically justified: no Pareto-suboptimal polymer appears in any minimum free-energy configuration, and the total equilibrium concentration of such polymers is small. We prove that there are finitely many Pareto-optimal polymers and exactly characterize them via a Hilbert basis computation, extending prior work from the saturated TBN model. To scale this approach to large systems, we develop a framework that restricts the number of different monomer types that a single polymer contains, and uses combinatorial covering designs to reduce the number of Hilbert basis computations required. We benchmark the method on several families of DNA molecular programming systems, demonstrating order-of-magnitude speedups over direct computation while recovering nearly all equilibrium-relevant polymers.
CommentsExtended version of paper accepted at DNA 32: 32nd International Conference on DNA Computing and Molecular Programming. Includes appendix
