Rethinking what pLDDT really tells us about protein flexibility

Deep-learning models have transformed structural biology by enabling reliable prediction of protein 3D structure models and providing confidence metrics such as predicted local distance difference test (pLDDT) to estimate local uncertainties. However, whether pLDDT reflects intrinsic protein flexibility remains unclear. Defining and quantifying flexibility and protein dynamics through experiments and computation is essential for advancing our ability to model and interpret conformational changes across different timescales.