Deep Learning-based Cancer Biomarker Discovery with Graph Neural Networks using Attention-based Knowledge Graph Embeddings and Protein-Protein Interactions.

GO-DMBC Model Architecture

Overview

This tool integrates multiple data sources to identify potential cancer biomarkers using protein-protein interaction networks, Gene Ontology annotations, and knowledge graph embeddings. The prediction model uses Deep learning architecture trained on cancer-specific datasets.

Supported Cancer Types

Breast Cancer

Glioblastoma

Lung Cancer

Analysis Pipeline

  1. Input: Gene list or pre-constructed PPI network
  2. PPI Construction: Build network from STRING database
  3. Embedding Generation: Generate GO term embeddings using fine-tuned BioBERT
  4. Feature Assembly: Concatenate GO and GeoKG embeddings
  5. Prediction: Deep learning model based classification
  6. Enrichment Analysis: GO and KEGG pathway analysis

Global Usage

Researchers using GO-DMBC worldwide

Start Analysis View Manual