Diversity indices and normalization approaches in microbiome studies

This thesis explores human microbiota, by investigating how to consistently measure biodiversity, both within and between different body sites, and how to correctly compare and normalize samples. In this work we propose a new method, called zero imputation, that tries to account for data sparsity; it proves to outperform common normalization approaches in recovering true data richness and relative abundances. We also develop a microbiome data simulator, useful to test all the methods reviewed