Detection of Crohn's disease using machine learning analysis of intestinal and blood B Cell receptor repertoire. CD biopsies BCR

Crohn disease (CD) is an autoimmune disease that attacks the gastrointestinal tract. Identification of CD may take many months and include colonoscopy and the use of abdominal capsule. B and T cell receptor repertoire sequencing has recently become a tool to monitor many aspects of the immune system. Gene usage, clones and amino acids in the variable regions of both B and T cell receptors have been used to identify many diseases including SARS-CoV2, celiac, HCV and others. In B cells, the process of somatic hypermutation (SHM) further contributes to receptor repertoire diversity within and between individuals. Yet SHM patterns have not previously been used to assess disease states. Here, we report CD-specific SHM patterns, revealed by a machine learning (ML) algorithm trained on B cell receptor repertoire sequences. These changes are found mostly in SHM hotspots introduced by DNA polymerase eta. The ML algorithm enables the prediction of CD with high accuracy.