Machine Learning Clustering for Cancer Analysis Employing Gene Expression Data

The idea that cancer types vary in their molecular structure (DNA, RNA, proteins, and epigenetics) depending on the origin and location of the cancer, has been worked on. The Cancer Genome Atlas (TCGA) has generated an initiative to carefully create a database to ensure quality data in the profiling of different tumors to promote research, a part of this large database was called Pan-Cancer, which has the genomic, epigenetic, transcriptional and proteomic profiling of 12 different types of cancer. In this research we took one of the profiling, RNA profiling, in 5 cancer types, in order to determine the possibility of segmenting in an unsupervised manner and to evaluate the difference of them by their origin. The results indicate that the number of clusters can vary from 5 to 7, with 5 clusters being established by the database labels, however, the division of 6 or 7 clusters is due to the clustering of breast cancer (BRCA) which has several origins.