Study design. A new data set of BCR DNA sequencing from intestinal biopsies of 25 CD patients as well as 25 controls was created. ML algorithms were trained on many whole repertoire parameters to classify CD. Results were compared with the ability to predict CD on a previously published RNA blood cohort. Only changes in SHM patterns succeeded in CD classification on both data sets. This figure was created with BioRender (https://biorender.com).