Perspectives of machine learning techniques in big data mining of cancer

Abstract

Advancements in cancer genomics and the emergence of personalized medicine hassle the need for decoding the genetic information obtained from various high-throughput techniques. Analysis and interpretation of the immense amount of data that gets produced from clinical samples is highly complicated and it remains as a great challenge. The future of cancer medical discoveries will mostly depend on our ability to process and analyze large genomic data sets by relating the profiles of the cancer genome to direct rational and personalized cancer therapeutics. Therefore, it necessitates the integrative approaches of big data mining to handle this large-scale genomic data, to deal with high complexity somatic genomic alterations in cancer genomes and to determine the etiology of a disease to determine drug targets. This demands the progression of robust methods in order to interrogate the functional process of various genes identified by different genomics efforts. This might be useful to understand the modern trends and strategies of the fast evolving cancer genomics research. In the recent years, parallel, incremental, and multi-view machine learning algorithms have been proposed. This chapter addresses the perspectives of machine learning algorithms in cancer genomics and gives an overview of state-of-the-art techniques in this field.