Conversational AI Agent for Precision Oncology: AI-HOPE-WNT Integrates Clinical and Genomic Data to Investigate WNT Pathway Dysregulation in Colorectal Cancer
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Introduction
The WNT signaling pathway plays a critical role in colorectal cancer (CRC) initiation and progression, particularly in early-onset cases among underserved populations. However, exploring WNT pathway alterations across clinical and genomic dimensions remains technically complex, limiting translational insights and personalized strategies. To address this, we developed AI-HOPE-WNT, a conversational artificial intelligence (AI) agent purpose-built for precision oncology. This system enables interactive, natural language querying of public cancer genomics datasets, specifically focusing on WNT pathway dysregulation in CRC.
Methods
AI-HOPE-WNT is a purpose-built conversational AI platform specifically designed to investigate dysregulation of the WNT signaling pathway in CRC. Developed using a modular architecture, the tool integrates large language models (LLMs), a natural language-to-code translation engine, and a backend statistical workflow that interfaces with harmonized CRC data from cBioPortal. Unlike general-purpose bioinformatics tools, AI-HOPE-WNT is optimized to support WNT-focused analyses, including integrative survival modeling, mutation frequency comparisons, odds ratio testing, and cohort stratification by clinical, genomic, and demographic variables. To demonstrate the utility of the platform, we replicated key findings from two of our prior studies examining WNT pathway alterations in high-risk CRC populations. These included survival analyses comparing WNT-altered and wild-type tumors across ethnicity and age subgroups, as well as mutation frequency analyses for key WNT genes such as RNF43 and AXIN2. Finally, we used AI-HOPE-WNT to generate novel hypotheses through exploratory queries on treatment response, mutation co-occurrence, and population-specific survival trends.
Results
In recapitulation analyses, AI-HOPE-WNT effectively reproduced key findings from prior studies, including: (1) improved survival outcomes associated with WNT pathway alterations in early-onset CRC and (2) a higher prevalence of RNF43 mutations among CRC patients from high-risk populations compared to lower-risk groups—both trends consistent with our previously published work. In exploratory mode, the platform identified several novel associations. Among early-onset CRC patients treated with FOLFOX, those harboring APC mutations exhibited significantly different survival outcomes compared to APC wild-type counterparts (p = 0.043). A separate analysis stratifying RNF43 -mutant tumors by stage revealed a significant survival disadvantage in metastatic cases relative to primary tumors (p = 0.028). Further, an investigation into AXIN1 and APC co-mutation patterns across tumor locations uncovered differential mutation enrichment and potential prognostic differences between colon and rectal adenocarcinomas. Notably, gender-stratified analyses among patients with AXIN2 mutations under varying microsatellite instability (MSI) statuses demonstrated significant survival variation (p = 0.036), suggesting a sex-specific molecular context. Lastly, in patients under 50 years of age, those with APC-mutated primary tumors had significantly worse overall survival (p = 0.031) and a higher odds of harboring APC mutations compared to their wild-type counterparts, underscoring the importance of age-stratified genomic analysis in early-onset CRC.
Conclusions
AI-HOPE-WNT is the first conversational AI agent specifically developed to investigate WNT signaling pathway dysregulation in CRC. This novel, accessible, and scalable platform enables natural language–driven analysis of integrated clinical and genomic data, transforming how researchers interrogate WNT-specific alterations across diverse patient populations. By automating complex bioinformatics workflows, AI-HOPE-WNT democratizes access to precision oncology tools, especially for non-programming users. Capitulation against published studies confirms its analytical rigor, while exploratory analyses demonstrate its unique capacity to uncover new associations related to treatment response, mutation co-occurrence, and health disparities. AI-HOPE-WNT establishes a new paradigm for pathway-specific precision medicine research and offers a powerful foundation for future biomarker discovery and targeted therapeutic strategies in CRC.
