This repository contains computational tools and metabolic models developed for the study "GAN-Enhanced Machine Learning and Metabolic Modeling Identify Reprogramming Signatures and Vulnerabilities in Pancreatic Cancer". This work is published in the PLOS Computational Biology journal. You could read this paper at: https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1013862
The framework integrates:
- Genome-scale metabolic models (GSMs) of 144 PDAC and 144 healthy samples
- WGAN-GP synthetic healthy data generation with biological validation
- Random Forest classifier for PDAC metabolic signature identification
- Multi-level analysis of pathway, reaction, and gene alterations