Large-scale cancer genomics data often comes in multiplatform and heterogeneous forms. These datasets impose great challenges in terms of the bioinformatics approach and computational algorithms. Numerous researchers have proposed to utilize this data to overcome several challenges, using classical machine learning algorithms as either the primary subject or a supporting element for cancer diagnosis and prognosis.
In this chapter, we will use some deep learning architectures for cancer type classification from a very-high-dimensional dataset curated from The Cancer Genome Atlas (TCGA). First, we will describe the dataset and perform some preprocessing such that the dataset can be fed to our networks. We will then see how to prepare our programming environment, before moving on to coding with an open source, deep learning library...
