As an essential technique for Graph Contrastive Learning (GCL), Graph Augmentation (GA) improves the generalization capability of the GCLs by introducing different forms of the same graph. To ensure information integrity, existing GA strategies have been designed to simultaneously process the two types of information available in graphs: node attributes and graph topology. Nonetheless, these strategies tend to augment the two types of graph information separately, ignoring their correlation, resulting in limited representation ability. To overcome this drawback, this paper proposes a novel GCL framework with a Joint spectrAl augMentation, named GCL-JAM. Motivated the equivalence between the graph learning objective on an attribute graph and the spectral clustering objective on the attribute-interpolated graph, the node attributes are first abstracted as another type of node to harmonize the node attributes and graph topology. The newly constructed graph is then utilized to perform spectral augmentation to capture the correlation during augmentation. Theoretically, the proposed joint spectral augmentation is proved to perturb more inter-class edges and noise attributes compared to separate augmentation methods. Extensive experiments on homophily and heterophily graphs validate the effectiveness and universality of GCL-JAM.

As a pivotal architecture in Self-Supervised Learning (SSL), Graph Contrastive Learning (GCL) has demonstrated substantial application value in scenarios with limited labeled nodes (samples). However, existing GCLs encounter critical issues in the graph augmentation and positive and negative sampling stemming from the lack of explicit supervision, which collectively restrict their efficiency and universality. On the one hand, the reliance on graph augmentations in existing GCLs can lead to increased training times and memory usage, while potentially compromising the semantic integrity. On the other hand, the difficulty in selecting TRUE positive and negative samples for GCLs limits their universality to both homophilic and heterophilic graphs. To address these drawbacks, this paper introduces a novel GCL framework called GRAph learning via Self-contraSt (GRASS). The core mechanism is node-attribute self-contrast, which specifically involves increasing the feature similarities between nodes and their included attributes while decreasing the similarities between nodes and their non-included attributes. Theoretically, the self-contrast mechanism implicitly ensures accurate node-node contrast by capturing high-hop co-inclusion relationships, thereby enabling GRASS to be universally applicable to graphs with varying degrees of homophily. Evaluations on diverse benchmark datasets demonstrate the universality and efficiency of GRASS. The dataset and code are available at URL: https://github.com/YukunCai/GRASS.

Background/Objectives: Integrating multi-omics data for cancer prognosis remains a challenging problem in bioinformatics because molecular profiles are high-dimensional, heterogeneous, and structured by incomplete biological relationships. Pathway databases provide biologically meaningful prior knowledge for modeling gene-level associations, but the sparsity and local incompleteness of pathway-derived networks often limit the performance of graph-based survival models. This study aimed to develop a pathway-guided framework for improving multi-omics survival prediction and identifying biologically relevant prognostic signals. Methods: We proposed PANA-Surv, a pathway-guided adaptive neighborhood augmentation framework for multi-omics cancer survival analysis. In this framework, KEGG pathways were used to construct gene graphs, and gene-level multi-omics profiles were encoded as node features. A conditional variational autoencoder module (PANA-VAE) was designed to enhance local representations through neighborhood reconstruction and adaptive weighting. The augmented features were then integrated into a graph convolutional survival model optimized with the Cox partial likelihood. Results: PANA-Surv was evaluated on 10 cancer cohorts from The Cancer Genome Atlas (TCGA). The proposed method achieved the highest mean concordance index (C-index) among all compared models and significantly outperformed Cox-EN, DeepSurv, GraphSurv, and LAGProg (all p < 0.01). Ablation analyses showed that both neighborhood reconstruction and adaptive weighting contributed to the observed performance gains, and KEGG-guided graph construction was more effective than alternative graph construction strategies. In a breast cancer (BRCA) case study, PANA-Surv identified 18 prognostic genes, including 12 genes supported by previous studies and 6 potentially novel candidates. Conclusions: These findings indicate that the integration of pathway prior knowledge with adaptive local feature enhancement can improve multi-omics survival modeling and support the identification of biologically relevant prognostic signals associated with cancer outcomes.