AI-HOPE-TP53: A Pathway-Centric AI Agent for Analyzing PI3K-Driven Molecular Events in Colorectal Cancer Precision Oncology
Abstract
Introduction
Early-onset colorectal cancer (EOCRC) is rising rapidly, especially among populations at risk who experience disproportionate incidence and mortality. The TP53 pathway, frequently altered in CRC, regulates key processes such as DNA repair and apoptosis. Despite its clinical relevance, TP53 dysregulation remains understudied in EOCRC, particularly in populations at risk. Current tools lack support for pathway-specific, population-stratified, and treatment-aware analysis. To address this gap, we developed AI-HOPE-TP53—the first conversational AI agent designed to explore TP53 alterations in CRC using harmonized clinical-genomic data and natural language queries.
Methods
AI-HOPE-TP53 leverages large language models (LLMs) to translate natural language prompts into executable bioinformatics workflows operating on datasets from cBioPortal. It supports dynamic cohort stratification by TP53-related genes, age, MSI status, treatment, tumor stage, and ethnicity. Automated analyses include mutation frequency, co-mutation profiling, survival curves, and odds ratios.
Results
AI-HOPE-TP53 successfully replicated known molecular characterization, identifying higher TP53 alteration rates in EOCRC patients. It revealed trends toward improved survival in patients with TP53 mutations and identified significant enrichment of H/L individuals among early-onset, FOLFOX-treated cases (OR = 2.002, p < 0.0001). Additional analyses uncovered subsite-specific outcomes in ATM-mutant CRC, stage-specific effects in CHEK1-mutated tumors, and disproportionate health burdens in treatment exposure.
Conclusions
AI-HOPE-TP53 enables fast, natural language–driven exploration of TP53 pathway alterations in CRC. Its ability to integrate clinical and genomic data across diverse subgroups makes it a powerful tool for precision oncology and Community-engaged health research—supporting both hypothesis validation and discovery in EOCRC and beyond.
Related articles
Related articles are currently not available for this article.