This workshop is practically full by now (18/03/2017), the organizers will have difficulties to accept new applications.
Having focused on simple model systems for two decades, molecular simulations of lipid membranes have recently started exploring more complex compositions and geometries, in an attempt to better account for the immense complexity of biological systems. What challenges lie ahead for molecular simulations of biological membranes? Which problems need to be solved during the next decade to make progress in this field? These questions are the focus of the CECAM workshop we are organizing.
Different sessions will be devoted to the future of force fields (both atomistic and coarse-grained), the challenges of rigorous, meaningful com-parisons between simulations and experimental measurements, and the challenges in simulat-ing complex systems (complex compositions, interaction with nano-sized particles of biological and synthetic origin, non-equilibrium effects).
1a. Atomistic and coarse grain force fields
The probably most essential ingredient of any simulation is the model, or “force field” (FF), and it is hence quite disconcerting that many properties extracted from biomembrane simulations depend quite sensitively on it. Moreover, the complexity of composition and the diversity of applications have substantially increased the demand for high-quality FFs, in particular concerning its transferability(1,2,3) and the diversity of observables compared to experiments or between different models(4) .
Paramount efforts have recently been made to parameterize FFs broad enough to embrace biological complexity(5), but even in ostensibly simple systems, many key properties of lipid membranes are still not well elucidated by simulations (lipid ordering(6), phase transition(7), electric potentials, etc.). FF improvement, and development of FFs based on different protocols, remains crucial(8,9,10).
1b. Linking experiment and simulation
Ultimately, both AA and CG FFs must be scrutinized against experiment. Unfortunately, our ability to accurately predict the outcome of experiments strongly depends on the observable, with local structural features (e.g., electron density profiles) being a lot easier than global colligative properties (e.g., bending rigidities).
While precise experimental measurements of, say, elastic parameters are still a major chal-lenge(11), we are finally reaching an era where numerical predictions become precise enough to highlight problems—either with the analysis strategy or with the underlying FF. This, in turn, affords us with cues for where to improve models(12,13,14).
Moreover, to deliberate the reliability of computational predictions, it is highly desirable to have conceptually independent means for accessing (what is believed to be) the same observable (for example, bending rigidities(15,16,17)). Such endeavors open new doors for confronting simulations with reality, and hence afford us with independent means for falsifying assumptions we hold dear, or improving models that have so far appeared acceptable.
Complexity (for instance in lipid compositions or membrane geometry) introduces new chal-lenges when performing and analyzing MD simulations. First of all, the number of relevant physical descriptors increases dramatically, leading to challenges concerning system equili-bration, as well as post-processing including the proper identification of the relevant degrees of freedom.
In addition, complexity may also prevent a straightforward comparison between simulations and experiments, due to the additional layer of interpretation that is needed to analyze both computational and experimental results(18,19,20). This raises fundamental questions that can unfortunately only be addressed on a case-by-case basis, namely the identification of the relevant parameters that need to be explicitly included in the model, as well as the reliability of experimental data that are used to build the model.