Potassium channels exhibit high selectivity and conductance, yet the atomic details of ion permeation, particularly the involvement of water molecules, remain debated. Two main conduction mechanisms have been proposed: the hard knock-on, in which ions traverse the selectivity filter in direct contact, and the soft knock-on, which involves co-permeation of water molecules. Using microsecond molecular dynamics simulations with the OPC water model, the AMBER19SB protein force field, and the 12-6-4 Sengupta et al. ion model, we observed that both hard and soft knock-on mechanisms are accessible and, notably, can reversibly transition in the MthK and KcsA channels across all simulated membrane potentials. These reversible transitions contrast with previous observations using the TIP3P water model, where water entry either disrupted conduction or was expelled, favoring exclusive hard knock-on events. Our results suggest that the choice of the water model, force field, and ion parameters significantly influences the observed conduction mechanism. Importantly, the coexistence of hard and soft knock-on in these simulations provides a potential reconciliation between structural data supporting hard knock-on and streaming potential measurements demonstrating water co-permeation. These findings reintroduce soft knock-on as a viable conduction mechanism and highlight the critical role of simulation parameters in reproducing potassium channel permeation behavior.
The Hidden Power of Normalization: Exponential Capacity Control in Deep Neural Networks
arXiv:2511.00958v1 Announce Type: cross Abstract: Normalization methods are fundamental components of modern deep neural networks (DNNs). Empirically, they are known to stabilize optimization dynamics and


