Fig. 1: An illustration of NMFGOT. | npj Biofilms and Microbiomes

Fig. 1: An illustration of NMFGOT.

From: NMFGOT: a multi-view learning framework for the microbiome and metabolome integrative analysis with optimal transport plan

Fig. 1

NMFGOT is designed for joint analysis of microbiome and metabolome data obtained from the same sample. a The microbial abundance and metabolite abundance profile matrices \({X}^{\left(1\right)}\) and \({X}^{\left(2\right)}\) are the inputs of NMFGOT (row represents microbial taxa or metabolite, and column represents sample). b By iteration fusion, NMFGOT learns the microbe loading matrix \({W}^{\left(1\right)}\), the metabolite loading matrix \({W}^{\left(2\right)}\), the clustering indicator matrices \({H}^{\left(1\right)}\) and \({H}^{\left(2\right)}\), and the sample-sample similarity matrix \(S\). c The enrichment analysis provides biological insights of the top metabolites identified in the matrix \({W}^{\left(2\right)}\). d Lasso penalized regression and stability selection are performed to identify significant and stable metabolite-microbe pairs, which provide an avenue to understand complicated disease mechanisms. e The sample-sample similarity matrix \(S\) facilitates visualization and data clustering.

Back to article page