Microbiome Forensic Database (MFDB) Classifier Training

This documentation explains how to train custom QIIME2 classifiers using the MFDB dataset (OTU/feature table + metadata on Zenodo) by filtering to a subset (e.g., body site/region) and re-training a classifier in QIIME2 (qiime2-amplicon-2024.10).

MFDB database:
https://mfdb.irm.uzh.ch/#/

Zenodo record (OTU table + metadata + trained classifier):
https://zenodo.org/records/18494561

Prediction workflow repository (optional):
https://github.com/meghnasw/Predict-mock-samples-forensic-BFID —

Documentation

• Data + artifacts (Zenodo): data
• Train your own classifier: training
• Visualize training outputs (.qzv): visualize
• Use a trained classifier to predict new samples: prediction
• Add more BioProjects: add-bioprojects
• Citation + contact: citation