A polarity-aware multi-relational model for the signed interaction prediction in biological networks
Researchers have developed a new deep graph model called PAMR (polarity-aware multi-relational model) to improve the prediction of signed interactions in biological networks. This model is specifically designed to differentiate between positive and negative interactions, which is crucial for drug discovery and repurposing. PAMR integrates graph convolutional networks with tensor decomposition and uses a conflict-aware sampling strategy to handle polarity ambiguities, outperforming existing baseline models in experimental evaluations. AI