Answers are grounded in this paper and cite their sources.
Motivation Existing approaches for inferring sample-specific protein interaction networks often rely on reference populations, evaluate interactions independently, and focus on single-omics data. Results We developed MOSSN, a random-walk framework that assigns sample-specific weights to reference protein interactions through within-sample molecular deviations and network propagation. Its multi-omics extension integrates gene expression, DNA methylation, and copy number variation. Benchmarking against six methods across 21,215 samples from 117 datasets spanning 18 cancer types showed the best average performance in tumor–normal discrimination, robustness, and survival prediction. Ablation analyses showed that within-sample molecular deviations and network propagation contributed to the performance. Multi-omics integration improved prognostic stratification compared with single-omics models. Compared with matched normal tissue, tumors across cancer types showed higher weights for interactions among cell-division and DNA-replication proteins and lower weights for metabolic and cytokine-signaling interactions. In immunotherapy-treated bladder cancer, average weights of interactions within immune-related genes were more strongly associated with response and survival than average expression of the same genes. MOSSN provides a single-sample framework for reweighting reference protein interaction networks that extends cancer studies from gene-centric to interaction-level representations. Availability and implementation MOSSN is available at https://github.com/bioUroZC/MOSSN.
No takes yet. Share an insight, caveat, or question.
Chen et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: