MIRROR: a miRNA regulation-level network-based algorithm to study sexual dimorphism in cancer

One of the open challenges in precision medicine, whose importance is growing every day, is sex-specific medicine: the study of how sex-based biological differences influence people’s health. These differences can be measured in terms of disease incidence, prevalence, mortality, and survival. Understanding the leading causes of these disparities is therefore of the utmost importance. With recent advancements in high-throughput technologies, large-scale molecular data are being generated for individual cancer patients; however, extracting meaningful insights from these complex datasets and translating them into clinical applications remains a challenge. Moreover, the functional interdependencies between the molecular components in a human cell often reflect the perturbations of a complex intracellular and intercellular network. Network-based approaches, being inherently holistic, can lead to a better understanding of the molecular mechanisms underlying a disease. For these reasons, this project focuses on the development of a network-based method to investigate sexual dimorphism in cancer using transcriptomic data. Many have already investigated transcriptomic data in this context, with particular interest in the role of miRNAs, showing the involvement of these regulatory elements in differentiating patients by sex in different types of cancer. However, these studies focus only on evaluating changes in the expression level, without conducting a more comprehensive analysis of miRNA expression and without investigating miRNAs’ targets. The aim of this project is therefore to carry out a multi- layer study involving both miRNAs and their target genes’ expression data. In particular, it focuses on the development of a novel and generalizable algorithm (MIRROR), which can be used on cancer patients to help identify key regulatory mechanisms and molecules that act as differentiators between males and females. Here we implemented and tested MIRROR on three different cancers (colon adenocarcinoma, hepatocellular carcinoma, and low-grade gliomas) and assessed its performance by comparing it to other state-of-the-art approaches. By doing so we proved MIRROR’s efficacy in identifying sex-specific key genes, presenting it as a viable alternative to the state-of-the-art methods which failed to capture these differences. Moreover, we also showed how the genes identified by MIRROR can be integrated with clinical features.