davesmith4398
commented
5 years ago Dear Samuli,
Thank you for your continuing interest and rich, substantive feedback. Our responses to your comments, many of which have involved edits and improvements to the document, are attached. I would particularly note that we have added you as a Contributor to the article. We thought that this was only fair, given your generosity of time and effort in the crucial, early stages of this document. The issue tracker mechanism will continue to be open, and we look forward to the future conversations with you and others.
Sincerely,
Dave Smith response_to_ollila_github_final.docx
davidlmobley
I have now red also the revised version with great interest. I have listed here some comments which came to my mind. Hopefully you find these useful.
1) Page 3: "For a detailed view of thermal fluctuations and the role of entropy" has been removed from page 3. Simulations can give very detailed information about fluctuations and entropy which are experimentally hard to access. Therefore, I think that this would be one of the key instances were simulations can give real added value. Thus, I would keep this point in the manuscript.
2) Page 3: "When the expected mesostructure of a lipid solution is not a planar bilayer but otherwise unclear" There are also other mesotructures than planar bilayer which can be well characterized by experiments and continuum theories, like hexagonal phases etc. However, if the mesostructure is something else than well known phase, like here http://dx.doi.org/10.1021/la404684r, the simulations can be useful. The sentence could be modified accordingly.
3) The list of example scientific questions has been added to page 3. I think that all these questions can be better answered experimentally than with current simulations, except these
It would be more useful to list practical examples with references where simulations give some solid and novel information which is not accessible with other techniques.
4) 3.3 Preparation of initial configurations "Lamellar lipid bilayers exist for a variety of concentrations of water to lipid ratios. All force fields will produce single bilayers if conditions are at full hydration (30 waters per lipid) or an order higher above." Large number of experimental conditions produce multilamellar phase. Most simulation systems are too small for the formation of real multilamellar phase. Therefore, it is not clear if simulations would form multilamellar, which they should, or a single bilayer. This sentence should be reformulated. Now it gives an impression that simulations have an artefact that they do not form multilamellar phases. We do not know if there is or not such artefact (if known, please cite).
5) Figure 1: Green and blue are very difficult to distinguish in the bottom.
6) Page 8: "As shown in “Statistical mechanics of inhomogeneous fluids”[114] by Schofield and Henderson, section IV, observables that can be cast as resulting from a global deformation of the system can be computed unambiguously from the profile, including curvature derivatives (see Sodt 2016[119], supplemental). Care must be taken, however, to interpret local features of the profile qualitatively. For example, it is appropriate to ask the question “Does the model capture the qualitative structure of the competing forces in lipid bilayer assembly and stability" This is quite technical and difficult to understand. I think that it would be better to clearly say that there is some potential ambiguity in the definition without too much details and cite appropriate publications for more details.
7) Page 9: "Computing error bars is tricky, but the order_parameters tool in LOOS [107] offers one approach" There is also one option presented in http://dx.doi.org/10.1021/acs.jpcb.5b04878
8) Page 9: "Since the area per lipid and NMR order parameter are tightly coupled" would be better as: "Since the area per lipid and NMR order parameter of acyl chains are coupled"
9) Page 9: "Calculated metrics can be compared with other simulations and directly with quadrupolar NMR splitting experiments [90]" would be better as "Calculated metrics can be compared with other simulations and directly with quadrupolar or dipolar NMR splitting experiments [90]"
10) Figure 3: It is not clear if these are values from simulations or experiments, and if from simulations, how realistic these are when compared to experiments.
11) Page 10, left column: State of the art comparison of bilayer dimensions, like area per molecule and thickness, to experiments is made using scattering forms factors, see e.g. http://dx.doi.org/10.1016/j.bbamem.2016.01.019 and references therein. I think that inclusion of form factors to the discussion would be highly relevant. Only the less accurate comparisons using lipid volume and order parameters are currently discussed.
12) Table 1: Different atomistic force fields have very different qualities for the headgroup and glycerol backbone structure (http://dx.doi.org/10.1021/acs.jpcb.5b04878) and cation binding affinity (http://dx.doi.org/10.1039/C6CP04883H). These informations should be included in the table. Some people also compare diffusion coefficients (http://www.cgmartini.nl/index.php/blog/269-jungle2, http://dx.doi.org/10.1021/acs.jpcb.6b01870)
13) Page 15: For the discussion about polarizable models, it could be interesting that the implicit inclusion of polarizability improves cation binding affinity to PC lipid bilayers: http://dx.doi.org/10.1021/acs.jpcb.7b12510
14) Page 21: "NMRlipids is a particularly new and comprehensive initiative launched by S. Ollila to compare membrane properties across 13 different force fields" The goal of the NMRlipids project is to "understand the atomistic resolution structures of lipid bilayers" (http://nmrlipids.blogspot.com/p/about.html). In the first publication there happens to be 13 models, however, the goal is not compare these but find the best possible models.