Hepatocellular Carcinoma tumor stage classification and gene selection using machine learning models.
Resumen
Cancer researchers are facing the opportunity to analyze and learn from big quantities of omic profiles of tumor samples. Different omic data is now available in several databases and the bioinformatics data analysis and interpretation are current bottlenecks. In this study somatic mutations and gene expression data from Hepatocellular carcinoma tumor samples are used to discriminate by Kernel Learning between tumor subtypes and early and late stages. This classification will allow medical doctors to establish an appropriate treatment according to the tumor stage. By building kernel machines we could discriminate both classes with an acceptable classification accuracy. Feature selection have been implemented to select the key genes which differential expression improves the separability between the samples of early and late stages.