生化網路的訊息流

Abstract

生物裡的生化網路有如電器裡的電子線路，決定著生命的進行方式。網路節點如何影響彼此造成訊息間的因果關係。此論文探討擁有適應性的生化網路拓樸裡的訊息流，分析網路節點間由內噪聲造成的轉移熵。由於該噪聲的共變異數可由Keizer的canonical theory決定，轉移熵可用Ito在高斯噪聲條件底下推導出的解析公式算出，計算結果比根據原始定義使用histogram統計的轉移熵還準。此結果讓我們能夠快速且精準地分析一般生物網路節點間的直接與間接因果關係。

A biochemical network in a living system is like an electric network in an electric device and decides the evolution of a life. How the nodes in the network affect one another leads to the causal relations between informations. This thesis is devoted to explore the information flow in the biochemical network topologies with good adaptation ability and to analyze the intrinsic noise induced transfer entropies between different nodes. Since the covariances of those intrinsic noises can be derived by Keizer’s canonical theory, the transfer entropy can be calculated by Ito’s analytical formula based on the condition of Gaussian noises. The calculated result is more precise than that evaluated from the histogram in the original definition of the transfer entropy. This result enables us to carry out a fast and precise analysis on the direct and indirect causal relations between different nodes in general biochemical networks.