Constructing the Microbial Association Network from Large-Scale Time Series Data Using Granger Causality

Genes

Dongmei Ai¹  LiC. Xia²  Gang Liu³  Xiaoxin Li³  Xiaoyi Liang³ 

[1] Basic Experimental of Natural Science, University of Science and Technology Beijing, Beijing 100083, China;Department of Medicine, Stanford University School of Medicine, 269 Campus Dr., Stanford, CA 94305, USA;School of Mathematics and Physics, University of Science and Technology Beijing, Beijing 100083, China;

关键词: Granger causality;    conditional Granger causality;    microbial association network;    time series data;    marine microbes;   

DOI  :  10.3390/genes10030216

来源: DOAJ

【 摘 要 】

The increasing availability of large-scale time series data allows the inference of microbial community dynamics by association network analysis. However, correlation-based association network analyses are noninformative of causal, mediating and time-dependent relationships between microbial community functional factors. To address this insufficiency, we introduced the Granger causality model to the analysis of a recent marine microbial time series dataset. We systematically constructed a directed acyclic network, representing both internal and external causal relationships among the microbial and environmental factors. We further optimized the network by removing false causal associations using the conditional Granger causality. The final network was visualized as a Granger graph, which was analyzed to identify causal relationships driven by key functional operators in the environment, such as Gammaproteobacteria, which was Granger caused by total organic nitrogen and primary production (p < 0.05 and Q < 0.05).

【 授权许可】

Unknown