Histone H3K4/H3K9 Methylation Dynamics Gate Cell Fate Plasticity In Pediatric High-Grade Gliomas: A Systems-Level Attractor Reconstruction

Background: This study employs computationally driven systems medicine approaches, integrating complexity theory and machine learning-driven discovery, to identify key biomarkers governing phenotypic plasticity in pediatric high-grade gliomas (pHGGs), specifically IDH-wildtype (IDHWT) glioblastoma and H3K27M Diffuse Midline Glioma (DMG). Methods: We integrate single-cell transcriptomics and histone mass cytometry to conceptualize pHGGs as complex adaptive ecosystems. Through attractor landscape reconstruction, network inference and algorithmic complexity measures, we computationally infer epigenetic and transcriptional regulators that steer cell fate cybernetics. Results: Our findings predict lineage-specific plasticity signatures, including KDM5B (JARID1B), ARID5B, GATA2/6, DLX5/6, FOXA2/FOXO1/FOXO3, ASCL4, ATF3, PRDM9, METTL5/8, RAP1B, DKK3, DOCK7, RLIM, TERF1, CD99, FABP5/7 and LAPTM5, alongside WNT, TGF[Formula: see text] and NOTCH pathway regulators, with subtype-specific and shared patterns across K27M and IDHWT tumors. We also identify endogenous bioelectric markers (GRIK3, GRIN3B, SLC5A9, NKAIN4, KCNJ4/6) as potential reprogramming targets. Previously reported plasticity signatures (PDGFRA, EGFR targets, OLIG1/2, FXYD5/6, MTSS1, SEZ6L, MTRNR2L1, SOX11) were independently recovered, supporting the robustness of our computational predictions. Collectively, results suggest that pHGG cell fates are biased toward neuronal-like identities, reflecting a teleonomic endpoint along pathological developmental attractors. Conclusions: By combining complex-systems modeling with translational AI, we predict in silico plasticity biomarkers and therapeutic targets for preventive, predictive and precision care. This work is computational-only and awaits experimental validation.