Understanding gene expression is essential in biology and its regulation is typically understood in terms of the binding of transcription factors, genomic sequence features, and modifications of the histone proteins; however, the exact relevance of the intrinsic three-dimensional nature of the chromatin filament is still poorly understood. Here we introduce Sequence-Enhanced Magnetic Polymer (SEMPER), a novel bottom-up polymer model that integrates data from various 1D biological assays. Through a novel Bayesian inference algorithm, we show quantitatively the importance of introducing polymeric constraints in statistical models that are otherwise physics-agnostic. The physically constrained model leads to better recapitulation of the epigenetic state of chromatin and the inference procedure yields plausible putative 3D structures without any input from chromatin conformation capture assays.
Ref.: A. Chen Yi Zhang, A. Rosa, G. Sanguinetti - Biophysical Journal 123, 184 (2024).