Developing a label propagation approach for cancer subtype identification problem

Abstract

The term of cancer is used to describe diseases in which abnormal cells that grow out of control and invade other tissues. There are multiple types of cancer and many types of cancer have various subtypes with different clinical and biological implications. These differences show that diverse methods should be followed for the treatment of different subtypes of cancer. Discovering cancer subtypes is an important problem in bioinformatics, as it can help improve personalized medicine. Knowing the subtype of cancer is useful for determine the treatment steps and prognosis. Computational bioinformatics methods help performing cancer analysis to design targeted treatments by exposing the common molecular pathology of different cancer subtypes. Thus far, several computational methods have been proposed to discover cancer subtypes or to stratify cancer into informative subtypes. However, existing works do not consider the sparseness of data, and result in ill-conditioned solution. To resort this shortcoming, in this thesis, we propose an alternative unsupervised computational method to stratify cancer into subtypes using applied numerical algebra techniques. More specifically, we applied this label propagation-based approach to stratify somatic mutation profiles of colon, head and neck, uterine, bladder and breast tumors. We then evaluated the performance of our method by comparing it to the baseline methods. Extensive experiments demonstrate that our approach highly renders tumor classification tasks by largely outperforming the state-of-the-art unsupervised and supervised approaches. tiplerinin