Identification and validation of microbial biomarkers from cross-cohort datasets using xMarkerFinder

Abstract Microbial signatures have emerged as promising biomarkers and targets for disease diagnosis, prognosis, and remission. However, these biomarkers exhibit contradictory results in different studies, necessitating the identification of universally robust microbial biomarkers. Therefore, we introduce xMarkerFinder, a four-stage computational framework for microbial biomarker identification with comprehensive validations from cross-cohort datasets, including differential signature identification, model construction, model validation, and biomarker interpretation. xMarkerFinder enables the identification and validation of reproducible biomarkers for cross-cohort studies, along with the establishment of classification models and potential microbiome-induced mechanisms. Although xMarkerFinder is initially developed for gut microbiome study, it is generalizable to different omics layers, as well as other habitats. Execution time varies depending on the sample size, selected algorithm, and computational resource. xMarkerFinder can be accessed at https://github.com/tjcadd2020/xMarkerFinder. FAQs, detailed tutorials, issue boards, and a ready-to-use Docker image are provided for users with no prior bioinformatics or statistics training.