Fig. 1
From: Compositional zero-inflated network estimation for microbiome data
Schematic illustration of COZINE. a Using OTU abundance data matrix as an input data, the transformed abundance matrix (\(\varvec{Y}\)) and incidence matrix (\(\varvec{V}\)) are generated. b Hurdle Model is fitted on the combined dataset of \(\varvec{Y}\) and \(\varvec{V}\) using neighborhood selection approach with group-lasso penalty. c Network structure is constructed based on the structure of zeros in the parameter set, \({\mathbf {G}}\), \({\mathbf {H}}\) and \({\mathbf {K}}\)