Deep learning has revolutionised machine learning with an unprecedented power to learn complex patterns. Big biological datasets, in particular multi-omics profiles from transcriptomes, epigenomics, and proteomes provide detailed descriptions of individual cells across entire organisms. Machine learning and deep learning algorithms have been extensively employed on these data to learn complex relationships between molecular characteristics of cells. These cellular patterns are regulated by molecular networks that can be presented by graph theory and modelled in neural network algorithms such as knowledge-primed neural networks (KPNNs; Fortelny and Bock, 2020).
Applications and further development of KPNNs requires extensive computational resources, such as those provided by the Mach2.
References