Advanced Tensor Decomposition-Based Integrated Analysis of Protein-Protein Interaction with Cancer Gene Expression Can Improve Coincidence with Clinical Labels

When cancer gene expression profiles obtained from The Cancer Genome Atlas (TCGA) are integrated with protein-protein interaction (PPI) information using tensor decomposition (TD), the coincidence between clinical labeling and latent variables provided by singular value decomposition (SVD) can increase, although PPI does not obviously contain information about clinical class outcomes. This suggests that TD has the power to extract hidden structure in PPI, which has biological meaning that can help gene expression better coincide with clinical information. From a practical standpoint, integrated analysis can also improve discrimination performance using class labels to a degree.