MULTIOMICS: HUMAN DISEASES AND MACHINE LEARNING

Abstract

Integrating different types of omics data with each other can help learn clinical-pathogenic changes that may be the cause of the disease, and these analyzes can be tested and confirmed by molecular studies in the next step. By integrating multi-omics algorithms together, it can map new relationships between molecules and disease agents, identify possible signaling pathways, and more comprehensively reveal biomarkers of disease. This approach, processing various omics data together, will facilitate revealing and detecting correlations between disease and genotype-phenotype and epigenetic factors. For example, it provides the advantage of molecular profiling in primary tissues via multi-omics to reveal the entire molecular mechanism of the course of the disease, and estimating the biological age of tissues and organs—in general, the entire organism—with a multi-omics algorithm to analyze the stages of aging step by step. In the past, single omics studies were performed in the hope of discovering the causes of pathologies and determining appropriate treatment. We now recognize that such an approach is far too simplistic. Many diseases simultaneously affect complex molecular pathways in which different levels of biological research interact. Therefore, there is a need for multi-omics studies that can include multiple omics studies simultaneously and provide more accurate maps of specific phenotypes/genotypes. Multi-omics approaches can be made more powerful by substructuring gene expression data, such as epigenomics. Complementary information can be obtained to better explain effects and causes, make and improve more accurate predictions, and understand molecular processes that are too complex to be understood through single-omics studies.