CMStatistics 2022: Start Registration
View Submission - CMStatistics
Title: Modelling heterogeneity in human gut microbiome: A clustering-based perspective Authors:  Angela Montanari - Universita di Bologna (Italy) [presenting]
Laura Anderlucci - University of Bologna (Italy)
Silvia Dallari - Alma mater studiorum- universita di Bologna (Italy)
Abstract: Microbiota are largely recognized as being central players in human health and in that of all organisms and ecosystems, and have been the subject of intense study. Next-generation sequencing techniques proved very effective for characterizing microbial communities by sequencing suitable molecular targets such as 16S ribosomal RNA gene amplicons for bacteria. However, the analysis and translation of microbiome data into meaningful biological insights remains very challenging. Firstly, microbiome data are compositional, i.e. microbial counts represent proportions instead of absolute abundances. Secondly, sparsity in the dataset can lead to false associations of microorganisms; a zero indicates either the absence of a microorganism, or an insufficient sequencing depth. Thirdly, it is challenging to differentiate between direct and indirect associations, in particular, if these are related to environmental factors. Different methods designed to account for unobserved heterogeneity are studied and compared on a real data set example.