CRC-Microbiome-Classification

The First International FeatureCloud Hackathon

20 - 24.06.2022, Hamburg, University of Hamburg, Institute for Computational Systems Biology

The microbiome challenge:

Data:

Shotgun metagenome data of gut microbiomes from ca. 1,400 colorectal cancer (CRC) patients acquired from ten distributed hospitals

Goal:

Prediction of colorectal cancer (CRC) phenotype (CRC, adenoma, healthy), and sub-phenotypes (e.g. CRC stage)

Additional difficulty:

Unbalanced class distribution, unbalanced features (e.g. gender, age), hidden confounders (we will simulate this with known features, e.g. age or BMI)

Overall objective:

A federated predictor of CRC (subtypes) from gut microbiome data

Validation:

Method performance will be compared to that of the standard centralized approach and will be validated by the organization team using new, unseen data

Director:

Prof. Dr. Jan Baumbach University of Hamburg Institute for Computational Systems Biology Notkestrasse 9 D-22607 HamburgGermany jan.baumbach@uni-hamburg.de

Reference:

https://featurecloud.ai/