MiMeNet

Exploring Microbiome-Metabolome Relationships using Neural Networks. MiMeNet predicts the metabolomic profile from microbiome data and learns undelrying relationships between the two. MiMeNet uses microbial features to predict metabolite output features. To do so, neural network hyper-parameters are first tuned. Then models are evaluated in a cross-validated fashion resulting in Spearman correlation coefficients (SCC) for each metabolite representing how well they could be predicted.