MarkerML – Marker Feature Identification in Metagenomic Datasets Using Interpretable Machine Learning

•Metagenomic marker features are conventionally identified by ‘class comparisons’ or through black-box machine learnt models.•MarkerML is developed to leverage interpretable machine learning for ‘class prediction’ based marker feature identification.•Dedicated webserver for metagenomics, it utilizes SHapley Additive exPlanations (SHAP) in companionship to hypothesis testing.•Minimal input data, provisions for pitfalls and intuitive workflows & visualizations enable convenient machine learning.•Hierarchical databases for microbial features (e.g. taxonomic lineage and pathway classes) aid feature characterization.