Multi-omics data integration for disease phenotype prediction using adversarial vision graph networks

Disease Phenotype Prediction is the process of predicting the manifestation of specific observable features related to a disease based on genetic data. The primary challenge is to predict disease phenotypes accurately by integrating heterogeneous, high-dimensional multi-omics data and environmental factors while considering complex interactions between genetic and environmental influences. Hence, an Adversarial Vision Graph Network with Red-Billed Blue Magpie Optimizer (Adv-VGN_RBMO) model is developed for disease phenotype prediction to allow for earlier detection of diseases based on genetic signatures, potentially improving patient outcomes. Here, the input Messenger Ribonucleic acid (mRNA) and Micro Ribonucleic acid (miRNA) expression data are accumulated, and then, the input mRNA and miRNA expression data are directed to the mRNA and miRNA interaction network. These interactions help reveal the pathways responsible for regulating gene activity. Moreover, both omics datasets and the results of their interaction network are provided as inputs to the phenotype prediction model. Here, the disease phenotype prediction is achieved by the Adv-VGN_RBMO model. The hyperparameters of Adv-VGN are tuned by Red-Billed Blue Magpie Optimizer (RBMO). Moreover, the Adv-VGN_RBMO recorded 95.988% recall, 95.168% accuracy and 94.768% precision with K-fold 8.