Discriminant and Network Analysis to Study Origin of Cancer

Enabled by rapid advances in biological data acquisition technologies and developments in computational methodologies, interdisciplinary research in machine learning for biomedicine tackles various challenging biological questions by comprehensively scrutinizing (multiplatform) data from multiple, distinct vantages. Understanding the origin and progression of cancer has great practical import for advancing both biological knowledge and potential clinical treatments. Technically, the most challenging biological questions inspire and promote the development and applications of novel computational methods. This chapter presents a coalition of state-of-the-art machine learning methods and leading-edge scientific puzzles. With DNA copy number and transcriptome data, we were able to design specific statistical hypothesis tests to reveal the origin of cancer by comparing the genomic and transcriptome codes and biological network structures.