Ulcerative colitis is a chronic inflammatory disease of the gastrointestinal system, affecting adults and children. Its cause is unknown, and the knowledge of reliable biomarkers is limited, especially for children. That makes the search for new biomarkers and pushing forth the analysis of the available data particularly challenging. We investigate proteomic data from children patients as a promising source, and tackle the problem implementing the recently developed parenclitic network approach to machine learning algorithms that solve classification task for proteomic data from healthy and diseased. We expect our approach to be applicable to other gastrointestinal diseases.