Computational chemistry and in particular, molecular simulation has been a key part of the chemist’s toolkit for decades. However, within the biological sciences, molecular simulations have been far slower in achieving wide-spread acceptance/adoption for a combination of reasons. I will discuss how we are using principles of computational chemistry to construct molecular representations of bacterial cell envelopes that account for as much of the molecular complexity as is feasible. I will discuss some of the unexpected phenomena we have observed directly as a consequence of accounting for molecular heterogeneity and realistic levels of crowding and how this has helped us generate new hypotheses. For example, possible mechanisms via which antimicrobial peptides move between the two membranes of the cell envelope. I will also describe the challenges we face in (i) ensuring such complex simulations are well-equilibrated and (ii) extracting the signals from the noise of vast array of molecular interactions. It is highly likely that deep learning methods will have a big role to play in the latter.
