Version française / Séminaires
Séminaire MODAL'X : Camille Champion (MAP5)
Publié le 19 septembre 2024
–
Mis à jour le 11 décembre 2024
Apprentissage statistique de réseaux microbiens, de l'inférence au clustering
Date(s)
le 12 décembre 2024
13h30-14h30
Lieu(x)
Résumé : Exploring how microbes interact within our intestines is a fascinating challenge in microbial ecology. Understanding these interactions helps us uncover the ways microbial communities influence both microbiota composition and the host system. New methods that reconstruct microbial network are published every year, but when different methods are used on the same data, they produce varying outcomes. There is currently no consensus on the best method to use. Our study introduces "OneNet" a new approach that combines seven methods to create a unified network. Our method showed on simulated data that it produced simpler networks while improving accuracy compared to the existing methods. On real gut microbiome data from liver cirrhosis patients, we identified in the network a cirrhotic cluster, composed of bacteria associated with degraded host clinical status. Our research contributes to unraveling the microbial interactions, potentially leading to better understanding of the role of the gut microbiota on health.
Mis à jour le 11 décembre 2024