Characterization of Tumor Composition Using Machine Learning

Wilms' tumor is the most common kidney cancer in children. Currently, it is thought that Wilms' tumor is caused by genetic and epigenetic distortions during various stages of fetal kidney development, leading to a great difference in the cellular composition of the tumor among patients. A group of researchers led by Prof. Tomer Kalisky from the Alexander Kofkin Faculty of Engineering at Bar Ilan University applied unsupervised machine learning for gene expression analysis of 53 tumors, taken from patients after chemotherapy treatment using three computational approaches. Their results demonstrated that each tumor is a unique mixture of three cell groups and that each tumor can be represented in the continuum of fetal kidney development. This work, carried out with computational biology tools, supports the relationship between Wilms' tumors and kidney development, paving the way for more quantitative tumor classification.

Read more >>