MPLasso: Inferring microbial association networks using prior microbial knowledge

Due to the recent advances in high-throughput sequencing technologies, it becomes possible to directly analyze microbial communities in human body and environment. To understand how microbial communities adapt, develop, and interact with the human body and the surrounding environment, one of the fundamental challenges is to infer the interactions among different microbes. However, due to the compositional and high-dimensional nature of microbial data, statistical inference cannot offer reliable results. Consequently, new approaches that can accurately and robustly estimate the associations (putative interactions) among microbes are needed to analyze such compositional and high-dimensional data. We propose a novel framework called Microbial Prior Lasso (MPLasso) which integrates graph learning algorithm with microbial co-occurrences and associations obtained from scientific literature by using automated text mining. We show that MPLasso outperforms existing models in terms of accuracy, microbial network recovery rate, and reproducibility. Furthermore, the association networks we obtain from the Human Microbiome Project datasets show credible results when compared against laboratory data.Author summary: Microbial communities exhibit rich dynamics including the way they adapt, develop, and interact with the human body and the surrounding environment. The associations among microbes can provide a solid foundation to model the interplay between the (host) human body and the microbial populations. However, due to the unique properties of compositional and high-dimensional nature of microbial data, standard statistical methods are likely to produce spurious results. Although several existing methods can estimate the associations among microbes under the sparsity assumption, they still have major difficulties to infer the associations among microbes given such high-dimensional data. To enhance the model accuracy on inferring microbial associations, we propose to integrate multiple levels of biological information by mining the co-occurrence patterns and interactions directly from large amount of scientific literature. We first show that our proposed method can outperform existing methods in synthetic experiments. Next, we obtain credible inference results from Human Microbiome Project datasets when compared against laboratory data. By creating a more accurate microbial association network, scientists in this field will be able to better focus their efforts when experimentally verifying microbial associations by eliminating the need to perform exhaustive searches on all possible pairs of associations.