Inferring metabolic networks from time series data
Other conference paper

香港中文大學研究人員

全文

其它資訊
摘要Metabolic networks comprise the chemical reactions of metabolism and the regulatory interactions that guide these reactions. Metabolic network behavior is commonly modeled by a set of differential equations that govern the time evolution of the metabolites. We focus on the description using biochemical system theory (BST) in the presence of noise��̇��=����Π��=1������������−����Π��=1������ℎ����+���� ,��=1,2,….,��Here ���� is the concentration of metabolite ��, ���� and ���� are respectively the overall rate constants of the processes that increase and decrease ���� , ������ and ℎ���� are parameters, known as kinetic orders, for such processes that involve metabolite ��, and ���� is a Gaussian white noise with ����(��)����(��′)= ��2��������(��−��′). We address the challenge of reconstructing such metabolic networks using only time series data ����(��). Such systems approach a stable fixed point ���� in the noise-free limit. In the presence of weak nose, by linearizing the dynamics about the fixed point, one can derive a relation����=���������� , ������=����Π��=1������ℎ����(������−ℎ����)/����,where ���� denotes the time-lagged covariance matrix of the measurements taken at two times separated by a time interval �� and ���� denotes the covariance matrix of measurements taken at the same time. Using glycolysis, the metabolic pathway that breaks down glucose to extract energy in living organisms, as an example we discuss how to use this relation to reconstruct the links of the network, namely to find which ������ and ℎ���� are nonzero, and the limitations of this method.
出版社接受日期12.10.2016
著者Emily S.C. Ching, Chumin Sun
會議名稱Dynamics Days Asia Pacific 9
會議開始日14.12.2016
會議完結日17.12.2016
會議地點HKBU, HKUST
會議國家/地區香港
會議論文集題名DDAP9 Programme Book
系列標題Programme Schedule – Posters
叢書冊次P20
出版年份2016
月份12
日期5
頁次11 - 11
語言美式英語

上次更新時間 2018-22-01 於 00:24