In some children, we can now trace a neurological condition to a single molecular cause — a mutation in a gene like SCN1A, which codes for an important sodium channel, or an autoantibody attacking a specific receptor. Yet the same fault can produce very different conditions, from easily controlled epilepsies to severe developmental syndromes such as Dravet syndrome. Much of that variability lies in the molecular detail: some mutations make channels open too readily, others stop them working at all.

The challenge is linking these smallest-scale disruptions to the whole-brain dysfunction they cause. We record what individual channels and synapses do, build computer models of neurons, develop patient-specific animal models, and "invert" models through dynamic causal modelling. These methods allow us to ask which molecular faults best explain a patient's brain dynamics. Across patients, animal models, and healthy volunteers, we trace the path from molecular fault to disrupted brain dynamics, working toward biomarkers, individual predictions, and better-targeted treatments.