This work used three different datasets. One for training the models and two to validate the prediction models.
We used American Gut Project (AGP) to train the model. Dataset was downloaded from AGP repository.
Two independent and external cohorts were used to validate our models.
- Chron´s subset from Gervers et al.
- Ulcerative colitis from Morgan et al.
Both datasets were retrieved from MLRepo repository created by KnightsLab. After clone MLRepo, you must to copy gevers
and morgan
folders into extdata
folder.