Machine learning-based feature selection to search stable microbial biomarkers: application to inflammatory bowel disease.

Taken together, our work not only showed how to improve biomarker discovery in the metataxonomic field without sacrificing classification performance but also provided useful insights for future comparative studies.