最新消息

2025

DriverOmicsNet 被Briefings in Bioinformatics 接受
  1. We propose DriverOmicsNet, a GNN-based model that integrates multi-omics data and gene co-expression networks to identify CDgs across 15 cancer types. 2. DriverOmicsNet incorporates both STRING PPI and WGCNA-derived co-expression networks to capture gene–gene interactions and co-regulatory structures, enhancing biological interpretability. 3. The model employs a GCN architecture to integrate diverse omics features, including gene expression, mutation, CNV, and DNA methylation. An autoencoder is used to compress methylation data for effective integration. 4. External validation using the CPTAC PAAD dataset demonstrates the robustness of DriverOmicsNet’s predictions. 5. Comparative analyses demonstrate that DriverOmicsNet achieves competitive performance relative to baseline models, with improved interpretability and biological relevance of identified CDgs.
Aug 15, 2025
DriverOmicsNet 被Briefings in Bioinformatics 接受
新文章發表於EClinicalMedicinee
Radiation dose escalation improves locoregional control and may enhance OS in SqCC, particularly in Asian populations, highlighting the need for histology- and region-specific therapeutic strategies. The choice of chemotherapy regimen may affect the interpretation of survival effects associated with high-dose. Furthermore, the genomic correlates of radiosensitivity and immune activation suggest potential for biologically guided dose personalization and combination with immunotherapy.
Aug 13, 2025
新文章發表於EClinicalMedicinee