AI in multi-omics analysis of COVID-19 patient data

The COVID-19 pandemic has led to an unprecedented increase in the volume of biological data generation, demonstrating the importance of developing an integrative and intelligent analytical framework. In the last few years, advancements in the artificial intelligence (AI) approaches have completely transformed the biological research landscape. Researchers have integrated the AI approaches with the multi-omics data generated from the COVID-19 patients to have a systems-level understanding of underlying disease mechanisms, predicting new variants and their spread rate, disease severity, immune response, and therapeutic opportunities. In this chapter, we have explored the utility of AI on multi-omics data. We started with an introduction to different kinds of omics data, such as genomics, epigenomics, transcriptomics, proteomics, and metabolomics. Next, we elaborated on what AI is and discussed its types, which include conventional machine learning methods (supervised and unsupervised), deep learning methods (autoencoders and convolutional neural networks), and network-based methods (graph neural networks, network propagation, and knowledge graphs). Next, we discussed different types of integration methods (early, intermediate, and late) used for integrating AI and multi-omics data. Moving ahead, we mentioned several applications of AI, such as biomarker discovery, host-pathogen interaction, drug repurposing, and predicting long COVID. Lastly, we mentioned several important projects and consortia and discussed several important case studies highlighting the usefulness of integrating AI with multi-omics data for personalized medicine.