#37 · Structural bio & protein design

Enhanced sampling and force-field corrections refine RNA and POLG conformational models

The common thread is computational access to conformational states that are rare, intermediate, or poorly constrained by a single static model. One preprint benchmarks enhanced-sampling MD (T-REMD, REST2, GaMD) and structure-prediction methods for RNA ensembles on HIV-1 TAR, validated by RDC and QM/MM comparisons. A second develops gHBfix-18Ab, an 18-component hydrogen-bond correction for AMBER RNA force fields that restores the NMR-observed stacked state of the MAPT A-bulge motif. A third compares AlphaFold2 models of POLG R627Q and W748S, placing the changes within model-variability range. What is new is the emphasis on controlling method error, through NMR-guided reparameterization and explicit variability controls. The set is heterogeneous: the first two address dynamic ensembles directly, while the POLG analysis concerns static prediction uncertainty.