Researchers Unveil MARRVEL-MCP: An AI Tool That Makes Genetic Interpretation as Easy as Asking a Question
"Is this BRCA1 mutation linked to cancer?" — A new computational tool can now answer that question by simply processing plain language, dramatically simplifying the complex task of genetic diagnosis.
A team of researchers at Baylor College of Medicine and Texas Children’s Hospital has developed MARRVEL-MCP, a groundbreaking computational tool that integrates large language models to help interpret genetic variants. The study detailing the tool was published in the American Journal of Human Genetics.
How MARRVEL-MCP Works
MARRVEL-MCP is a significant upgrade to the existing MARRVEL platform, which already aggregates data from genomic, functional, and model-organism databases.
The new tool allows users to input plain language questions, such as "Is this BRCA1 mutation linked to cancer?" The system then automatically identifies relevant information, queries multiple biological databases, and produces answers based on available evidence.
This approach is designed to reduce the time and specialized expertise typically required to analyze variant data, making genetic diagnosis far more accessible to non-experts.
Performance and Accuracy
In a key demonstration of its capability, the researchers reported that when a smaller AI model (gpt-oss-20b) was integrated with MARRVEL-MCP, its accuracy improved dramatically from 41% to 94%.
"Not all genetic changes linked to a condition are necessarily disease-causing," said Dr. Hyun-Hwan Jeong, co-corresponding author, assistant professor of pediatrics – neurology at Baylor, and investigator at the Duncan Neurological Research Institute at Texas Children's. He added that identifying which variants contribute to disease is a crucial but complex and time-consuming process.
Background and Innovation
The original MARRVEL platform allowed simultaneous searches of multiple biological databases, but it required precise input and manual interpretation of complex outputs. The developers stated that MARRVEL-MCP addresses these limitations by incorporating AI agents that autonomously execute multi-step analytical workflows from simple language queries.
Availability
MARRVEL-MCP is available as an open resource at: https://chat.marrvel.org
Contributors and Support
First author Zachary Everton, along with Jorge Botas, Seon Young Kim, and Lin Yao, contributed to the work. The study was supported by funding from:
- The Cancer Prevention and Research Institute of Texas
- The Chan Zuckerberg Initiative
- The National Institutes of Health
- Other sources