Bayesian reweighting of α-synuclein ensembles reveals local-global discrepancies between chemical shifts and chain dimensions.

α-Synuclein is an intrinsically disordered protein whose ensemble is shaped by both local structural preferences and global chain dimensions. Here, we combined BioMatics/AlphaFold2/AFflecto-derived structures with molecular dynamics simulations to generate a pool of α-synuclein conformations. Predicted chemical shifts were reweighted against experimental NMR chemical shifts using Bayesian Maximum Entropy reweighting. Global analyses showed that MD conformations remained compact and internally contact rich, while BioMatics/AlphaFold2/AFflecto conformations occupied a more expanded Rg and Ree regime. These results reveal a local-global discrepancy in α-synuclein ensemble refinement. Namely, chemical shift-guided BME preferentially supports MD-derived local conformations, whereas global descriptors favor a more expanded BioMatics/AlphaFold2/AFflecto-like ensemble. Our findings highlight the need to combine local NMR restraints with global observables such as SAXS, smFRET, and PRE for accurate modeling of intrinsically disordered protein ensembles. A major future improvement would be to develop multi-restraint BME reweighting, where chemical shifts are combined with SAXS-derived Rg, smFRET distances, PRE-derived long-range contacts, RDCs, hydrodynamic radius, and NMR relaxation data.