> Read data were downloaded from the National Center for Biotechnology Information’s (NCBI’s) Sequence Read Archive (SRA), including data from the Human Microbi

> Read data were downloaded from the National Center for Biotechnology Information’s (NCBI’s) Sequence Read Archive (SRA), including data from the Human Microbiome Project (34) and several other studies (25 Bioprojects in total) pursuing massively parallel sequencing of human metagenomic samples Don't you need a statistical model accounting for possible sequencing batch effects, in that case? They may have been batched according to some scheme which correlates with the chronic diseases, and variations in the sequencing chemistry between batches could cause spurious associations.