Leibovich N
National Research Council of Canada, NRC-Fields Mathematical Sciences Collaboration Centre, 222 College st., Toronto, ON, M5T 3J1, Canada.
Sci Rep. 2025 Jan 23;15(1):3004. doi: 10.1038/s41598-025-86332-0.
Revealing interactions in complex systems from observed collective dynamics constitutes a fundamental inverse problem in science. Some methods may reveal undirected network topology, e.g., using node-node correlation. Yet, the direction of the interaction, thus a causal inference, remains to be determined - especially in steady-state observations. We introduce a method to infer the directionality within this network only from a "snapshot" of the abundances of the relevant molecules. We examine the validity of the approach for different properties of the system and the data recorded, such as the molecule's level variability, the effect of sampling and measurement errors. Simulations suggest that the given approach successfully infer the reaction rates in various cases.
从观测到的集体动力学中揭示复杂系统中的相互作用是科学中的一个基本逆问题。一些方法可能会揭示无向网络拓扑结构，例如使用节点 - 节点相关性。然而，相互作用的方向，即因果推断，仍有待确定——尤其是在稳态观测中。我们介绍了一种仅从相关分子丰度的“快照”来推断该网络中方向性的方法。我们研究了该方法对于系统的不同特性以及所记录数据的有效性，例如分子水平的可变性、采样和测量误差的影响。模拟表明，给定的方法在各种情况下都能成功推断反应速率。
