Understanding structure-function relationships in allosteric systems beyond the linear regime

Proteins use allostery to regulate activity at active sites via a long-ranged mechanical response. In the past, elastic network models have been trained to perform allosteric tasks using principles of physical learning. It is known that (1) the allosteric mechanism is non-reciprocal and (2) allostery takes place beyond the linear regime. We extend previous work to understand structure-function relationships in systems beyond their linear response regime. We train elastic networks to perform more realistic allosteric tasks and examine the structure-function relationship in these cases.