Artificial Intelligence in the prediction of 3D chromatin structure and gene regulation

All human cells share the same DNA sequence, yet their functional diversity is largely determined by the three-dimensional (3D) organization of chromatin within the nucleus. This spatial architecture regulates gene accessibility and expression, linking linear genomic information to cellular function. Recent advances in chromosome conformation capture technologies, such as Hi-C, ChIA-PET or HiChIP, have revealed hierarchical chromatin structures including compartments, topologically associating domains (TADs) and loops. Traditional computational methods often struggle to model these complex patterns, but artificial intelligence (AI), particularly deep learning, has emerged as a powerful tool for predicting both gene regulation and 3D chromatin organization from sequence and epigenomic features. Integrative and multimodal approaches, extended by self-supervised models, enable accurate modeling of regulatory mechanisms across cell types, providing novel insights into genome function and the molecular basis of disease.