The purpose of our work is to understand how channels and other proteins move between conformations, how large motions and allosteric modulation are induced by small external factors (e.g. ligand binding, changes in membrane potential, or lipid composition), and to develop techniques to combine several sources of experimental data such as cryo-EM, tomography and electrophysiology with bioinformatics and simulation to create models of the entire ensemble of structural and functional states for ion channels and transporters. Our computational work supported by NAISS enables us to build atomic-resolution models to study features not readily available in experiments, such as transitions between states and finding new ways to use low-resolution experimental data, in particular to model structural biology beyond individual proteins. Our primary model system is the pentameric ligand-gated ion channels responsible for mediating chemical singling across the synapse in the nervous system, as well as their bacterial homologs. While we increasingly have structural data for these channels, we still do not understand the molecular mechanism of gating, and the desensitised state all channels end up in shortly after activation is still almost entirely uncharacterised. We seek to use computational methods to sample motions based on experimental data, in particular combining molecular dynamics simulations in combination with Markov State Models to determine both the free energy landscape of channel activation to systematically investigate how the composition of subunits and allosteric modulators cause the fascinating diversity of the central nervous system activation. A second critical application is that we depend on large scale computational resources to reconstruct high-resolution structures of proteins, and increasingly also tomographic data. Our second key target system is TRPM3, tetrameric non-selective cation channel and member of the transient receptor potential superfamily. It mediates noxious heat sensation and inflammatory heat hyperalgesia, and its hyperactivity leads to pain, and it is a promising target for the treatment of pain. Drug modulation of TRPM3 is, however, still poorly understood, and we are combining cryo-EM structures with simulations to resolve this. We combine both our MSM sampling and density fitting codes with new conformations predicted from AlphaFold, based on the idea that AF2 frequently generates a whole range of candidate clusters - and at least in a few cases lower-ranking clusters appear to have corresponded to alternative conformations, which are enabling us to model conformational transitions without having to resort to classical molecular dynamics simulations. An important part of our work is that we are able to combine cryo-EM data with molecular dynamics, which has enabled several studies with deep mechanistic insight about gating and modulation. For the present project, we are also working with sampling and generative AI methods to attempt to identify new - transient - pockets in protein targets that have previously been considered undruggable.