Dango: Predicting higher-order genetic interactions (opens in new tab)

Zhang et al. present DANGO, a computational framework utilizing a self-attention hypergraph neural network to model higher-order genetic interactions. By integrating multimodal molecular networks and protein sequence embeddings, DANGO expands the quantitative yeast trigenic interaction landscape to over 400 million interactions, uncovers functional genetic modules, and improves genotype-to-phenotype mapping across diverse environmental conditions.