Parenclitic hypergraphs and their application in personalized cancer therapy

Abstract

Understanding the differences between individual instances of the same complex system remains a central challenge, particularly in biological contexts. Parenclitic networks constitute a suitable means to detect deviations in correlations with respect to reference populations. Here, we introduce parenclitic hypergraphs, a general framework for identifying anomalies in higher-order correlations across arbitrary interaction orders. After validating the method on synthetic datasets and benchmark ones, we apply it to patient-derived cancer organoids, capturing temporal changes in gene expression between healthy and cancerous tissues as the disease progresses. Our approach not only reproduces known oncogenic signatures, but also reveals a previously unrecognized candidate therapeutic target. Since organoids are generated from individual patients, our method provides, for the first time, a viable protocol for personalized cancer therapy based on higher-order correlation patterns. These findings offer a novel, systems-level strategy for precision oncology grounded in complex systems theory.