Researchers at the University of Navarra in Spain have launched RNACOREX, an open-source software that uncovers hidden genetic networks in cancer tumors. The tool analyzes thousands of molecular interactions and predicts patient survival with clarity rivaling advanced AI systems. Tested on data from 13 cancer types, it provides interpretable insights to advance cancer research.
Researchers from the University of Navarra have introduced RNACOREX, a new open-source platform designed to reveal the complex genetic networks underlying cancer. Developed at the Institute of Data Science and Artificial Intelligence (DATAI) in collaboration with the Cancer Center Clínica Universidad de Navarra, the software integrates data from international biological databases with gene expression information to identify key miRNA-mRNA interactions.
These interactions form regulatory networks that influence tumor behavior and patient outcomes. As Rubén Armañanzas, head of the Digital Medicine Laboratory at DATAI and a lead author, explains, "Understanding the architecture of these networks is crucial for detecting, studying, and classifying different tumor types. However, reliably identifying these networks is a challenge due to the vast amount of available data, the presence of many false signals, and the lack of accessible and precise tools capable of distinguishing which molecular interactions are truly associated with each disease."
RNACOREX addresses these issues by building probabilistic models from ranked interactions, offering a clear molecular map of tumor function. Evaluated using data from The Cancer Genome Atlas (TCGA) consortium, the tool was applied to 13 cancer types, including breast, colon, lung, stomach, melanoma, and head and neck tumors. It predicts survival with accuracy comparable to sophisticated AI models but stands out for its interpretable explanations.
Aitor Oviedo-Madrid, first author and a researcher at DATAI, notes, "The software predicted patient survival with accuracy on par with sophisticated AI models, but with something many of those systems lack: clear, interpretable explanations of the molecular interactions behind the results." Beyond predictions, RNACOREX identifies shared molecular patterns across tumors and highlights biomedically relevant molecules, potentially aiding in new hypotheses for diagnostics and treatments. Oviedo-Madrid adds, "Our tool provides a reliable molecular 'map' that helps prioritize new biological targets, speeding up cancer research."
Published in PLOS Computational Biology in 2025 (DOI: 10.1371/journal.pcbi.1013660), RNACOREX is available on GitHub and PyPI, with automated database tools for easy integration. Funded partly by the Government of Navarra's ANDIA 2021 program and ERA PerMed JTC2022 PORTRAIT, the project emphasizes explainable AI in genomics. Armañanzas states, "As artificial intelligence in genomics accelerates, RNACOREX positions itself as an explainable, easy-to-interpret solution and an alternative to 'black-box' models, helping bring omics data into biomedical practice."
Future expansions include pathway analysis and additional molecular layers to better model tumor progression, supporting precision cancer medicine.
