NetG2P: Network-based genotype-to-phenotype transformation identifies key signaling crosstalk for prognosis in pan-cancer study
AI-Powered Insights for Precision Oncology in Oncology
This study introduces NetG2P, a novel network-based framework that integrates genomic and expression data with network propagation and explainable AI to identify Critical Oncogenic Features (COFs) and their signaling crosstalk for cancer prognosis. Applied to TCGA data across 10 cancer types, NetG2P effectively stratifies patients into short- and long-term risk groups. The framework reveals distinct patterns of pathway interaction networks (distributed vs. modular) among cancer types, correlating with cancer hallmarks and GO enrichments. Furthermore, NetG2P successfully predicts novel drug targets and identifies drug repurposing candidates for high-risk patient groups using cancer cell line perturbation data. This approach represents a significant step towards personalized treatment strategies in precision oncology.
Quantifiable Impact: NetG2P's Proven Results
NetG2P delivers tangible improvements in cancer prognosis prediction and personalized treatment identification, translating complex genomic data into actionable insights for precision oncology.
0.668
Average prediction accuracy across 12 cancer types, higher than other models (0.262-0.393)
60,660
Genes reduced to 1,485 oncogenic features, simplifying ML models
617
Unique COFs identified across 10 cancer types, with 395 shared and 222 cancer-specific
8
Compounds identified as more effective for short-term risk groups, e.g., Fulvestrant, Olaparib
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
NetG2P Overview
NetG2P is a comprehensive framework that transforms patient-specific somatic mutation profiles into functional states of oncogenic pathways and their signaling crosstalk. It uses network propagation to integrate genomic and expression data, then applies machine learning and explainable AI to identify critical oncogenic features (COFs) associated with cancer prognosis. These COFs are then used to stratify patients into risk groups and suggest personalized treatments.
Signaling Crosstalk Explained
Signaling crosstalk involves shared components between two or more biological pathways, playing a crucial role in signal integration and pathway reactivation. NetG2P specifically defines pathway crosstalk as the overlap gene sets between KEGG pathways, generating 644 non-empty pathway-link features from 54 oncogenic pathways. Understanding these crosstalks is essential for developing effective combination therapies.
Pan-Cancer Insights
The pan-cancer analysis revealed distinct structural patterns of pathway interaction networks among different cancer types. Bladder (BLCA), ovarian (OV), and uterine corpus endometrial carcinoma (UCEC) exhibited a 'distributed' network, affecting multiple cancer hallmarks. In contrast, stomach (STAD) and liver (LIHC) cancers displayed a 'modular' network, influencing one or two hallmarks. This suggests that the underlying architecture of pathway interactions is crucial for determining cancer behavior.
Drug Repurposing Potential
NetG2P leverages cancer cell line perturbation data (DepMap, GDSC) to predict novel drug targets and repurposing candidates for high-risk patient groups. By comparing dependency scores and IC50 values between short- and long-term risk cell lines, the framework identified genes within COFs (e.g., WNT5B, WNT11 in STAD) and specific compounds (e.g., Axitinib, Fulvestrant, Olaparib) that show differential effectiveness, facilitating personalized treatment strategies.
0.668 MCC
Average prediction accuracy across 12 cancer types, outperforming other models.
NetG2P Workflow
Patient Genomic Data (SNV + Expression)
→
Network Propagation (PPIN)
→
Oncogenic Feature Matrix (Pathways & Crosstalk)
→
Machine Learning & XAI
→
Critical Oncogenic Feature (COF) Identification
→
Risk Stratification & Drug Targets
| Feature Set | Traditional Omics (Genes) | NetG2P (COFs) |
|---|---|---|
| Input Data Complexity |
|
|
| Interpretability |
|
|
| Prediction Performance (MCC) |
|
|
| Identification of Crosstalk |
|
|
Impact on Stomach Adenocarcinoma (STAD) & Liver Hepatocellular Carcinoma (LIHC)
For STAD and LIHC, NetG2P identified a 'modular' pathway interaction network structure, implying that mutational impacts are localized within modules before spreading. This distinct pattern correlated with high enrichment in specific cancer hallmarks like enabling replicative immortality (STAD) and sustaining proliferative signaling (LIHC), but not across multiple hallmarks as seen in 'distributed' networks. Furthermore, in STAD, NetG2P highlighted Wnt and mTOR pathway crosstalk (P10 and P9) as key long-term COFs and potential drug targets for drug repurposing candidates like Axitinib, which modulates Wnt pathways. This exemplifies how NetG2P's structural analysis can guide targeted therapeutic strategies.
Highlight: Modular networks enable targeted interventions.
Calculate Your Potential ROI with AI
See how NetG2P-like AI solutions can drive efficiency and significant cost savings within your organization. Adjust the parameters below to estimate your return on investment.
Estimated Annual Savings
$520,000
Annual Hours Reclaimed
10,400
Our Proven Implementation Roadmap
We guide enterprises through a structured process to integrate advanced AI solutions, ensuring seamless adoption and measurable success.
Phase 1: Discovery & Strategy
Comprehensive assessment of your current data infrastructure, business objectives, and integration points to tailor the NetG2P framework to your specific needs.
Phase 2: Data Integration & Model Training
Securely integrate your genomic and clinical data. Our experts will preprocess, clean, and use this data to train and fine-tune NetG2P's predictive models.
Phase 3: Validation & Insight Generation
Rigorously validate model performance with your specific datasets. Generate critical oncogenic features and initial prognostic insights for your patient cohorts.
Phase 4: Clinical & Therapeutic Application
Translate AI-driven insights into actionable clinical strategies, identifying patient risk groups and proposing novel drug targets or repurposing candidates.
Phase 5: Continuous Optimization & Support
Provide ongoing monitoring, model updates, and dedicated support to ensure sustained performance and adaptation to evolving clinical and genomic data.
Ready to Transform Your Oncology Research?
Discover how our AI solutions can accelerate drug discovery, enhance prognostic accuracy, and pave the way for true precision medicine. Let's discuss a tailored strategy for your enterprise.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat