Thanks, Sander! 😄

Thanks! The predicted load can be used to make the profile less compositional by multiplying it with relative abundances. However, in the disease-related analysis, the impact of such normalization seemed to be small, except for extreme diseases such as IBD.

Huge thanks to all collaborators and consortium members for their substantial contributions to the study! @borklab.bsky.social , @GALAXY, @MicrobLiver,

We thank all the collaborators and consortium members who were involved in the study!:
#GALAXY, #MicrobLiver, @borklab.bsky.social

Our prediction tool, named Microbial Load Predictor (MLP), is freely available from here. microbiome-tools.embl.de/mlp/

Importantly, we found that microbial signatures of each disease are strongly correlated with the altered microbial load of patients, indicating that change in microbial load, rather than the disease itself, is the main driver in shaping altered microbial profile in patients.

By applying our model to global metagenomes from previous studies (n = 34,539), we revealed a lot of novel associations between predicted microbial loads and host factors such as age, sex, diet, disease, and medication.

Metagenomic analysis reveals only relative abundances of microbes but quantification of absolute abundances requires laborious experiments. We developed a novel tool to quantify fecal microbial load (i.e. total microbial cells per gram) purely based on relative species profiles.