Expression Analysis of MicroRNA Isoforms in Molecular Subtypes of Breast Cancer

During biogenesis, miRNA genes are transcribed by RNA polymerase II, forming the primary miRNA, which undergoes equivalence with the Drosha enzyme. Dicer then processes the molecules to form pre-miRNA. At the 5' end of the miRNA there is a special region (seed region) at positions 2–7, complementary to the target mRNA. As part of the miRISC complex, microRNA interacts with mRNA and causes its degradation directly or by attracting other proteins. The alternative separation of Drosha and Dicer results in the formation of microRNA isoforms (isomiRs) with different 5′ and/or 3′ ends. Non-template modifications are also sources of isoforms. It has been shown that information about the presence or absence of a microRNA isoform allows the existence of 32 types of tumors. It was also found that the expression pattern of isomiRs can better distinguish a tumor from normal breast tissue than the pattern of microRNAs. IsomiR is also capable of detecting and molecularly characterizing the luminal A and luminal subtypes. During our work, we analyzed 1186 samples from the TCGA-BRCA project. Using Recursive Feature Elimination, isomiR was selected, which allows cells to be classified into molecular subtypes of the mammary gland. A classifier was also trained and a classification was obtained. In the course of our work, we found that the expression pattern of microRNA isoforms allows for good separation of proteins (AUC-ROC = 0.96) into molecular subtypes, as well as identification of healthy tissues.