davesmith4398
commented
6 years ago Hi Samuli,
Thank you so much for your thorough read of this manuscript and highly beneficial feedback. We are in the process of addressing your comments and suggestions, and are awaiting further feedback from the peer review process. Once we receive those reviews, we intend to compile all edits and responses and address your GitHub Issues directly to hopefully arrive at a consensus for a refined and improved Best Practices document. Thank you again for your time and efforts, and we will be in touch.
Sincerely,
Dave Smith
I have red your manuscript with great interest and I have several comments and suggestions. I will make separate issues on few important topics that I have in mind, but I will start here by suggesting some recent review articles which you, and the targeted audience of the manuscript may find useful. I also list several small comments and suggestions below.
POTENTIALLY INTERESTING LITERATURE
Review discussing practical aspects of running lipid bilayer MD simulations:
Efficient preparation and analysis of membrane and membrane protein systems Matti Javanainen and Hector Martinez-Seara Biochimica et Biophysica Acta (BBA) - Biomembranes Volume 1858, Issue 10, October 2016, Pages 2468-2482 https://doi.org/10.1016/j.bbamem.2016.02.036
Review discussing combining lipid bilayer MD simulations with NMR and scattering data: Atomistic resolution structure and dynamics of lipid bilayers in simulations and experiments O.H. Samuli Ollila and Georg Pabst Biochimica et Biophysica Acta (BBA) - Biomembranes Volume 1858, Issue 10, October 2016, Pages 2512-2528 https://doi.org/10.1016/j.bbamem.2016.01.019
Review discussing different methods to calculate elastic constants from simulations: The Membrane Bending Modulus in Experiments and Simulations: A Puzzling Picture D. Bochicchio and L. Monticelli Advances in Biomembranes and Lipid Self-Assembly Volume 23, 2016, Pages 117-143 http://dx.doi.org/10.1016/bs.abl.2016.01.003
Publications comparing performance of different force fields against experiments:
A Critical Comparison of Biomembrane Force Fields: Structure and Dynamics of Model DMPC, POPC, and POPE Bilayers Pluhackova et al. DOI: 10.1021/acs.jpcb.6b01870 J. Phys. Chem. B 2016, 120, 3888−3903
Molecular electrometer and binding of cations to phospholipid bilayers Catte et al. Phys. Chem. Chem. Phys., 2016,18, 32560-32569 DOI: 10.1039/C6CP04883H
Toward Atomistic Resolution Structure of Phosphatidylcholine Headgroup and Glycerol Backbone at Different Ambient Conditions Botan et al. J. Phys. Chem. B, 2015, 119 (49), pp 15075–15088 DOI: 10.1021/acs.jpcb.5b04878
COMMENTS AND SUGGESTIONS:
The references are not listed in the order of the appeareance.
Page 2, left column, second paragraph "Lamellar lipid bilayers often exist as vesicles (also known as liposomes–basically bilayer spheres) in solution, but can also exist as planar bilayers in periodic simulation systems, with experimental supports, or multilayer stacks for struc- tural analysis." This could be little bit reformulated because multilamellar stack, not unilamellar vesicles, is the thermodynamically stable phase for most lipids that form lamellar phases.
Page 2, left column, end of the second paragraph I would not call bilayer as a Gibbs monolayer. As far as I know, Gibbs monolayer is usually used for "a monolayer formed by a compound that is soluble in one of the phases separated by the interface on which the monolayer is formed. (wikipedia)" Such a monolayer has a surface pressure, which depends on the chemical potential of the compounds. I think that this is different from a bilayer.
Page 2, right column, bulleting point list I have hard time to understand what does this mean: "When bilayer self-assembly and the saturation assump- tion are not a given"
Page 3, right column "In general, force field developers seek to first capture thermodynamic properties, then address dynamical properties." I think that also structural properties are also substantially addressed.
Page 4, right column, third line "However, force fields like CHARMM36 [42] appear to be well-tuned for phase changes within 5 o C." I would be happy to see reference for this. Pluhackova et al. [ J. Phys. Chem. B 2016, 120, 3888−3903 DOI: 10.1021/acs.jpcb.6b01870 ] seems to argue that the phase transition temperature would be too high.
Page 4, left column, before equatio (1) frame tension is not defined.
Page 4, Equation (1) L_z, P_z, P_x and P_y are not defined.
Section 3.3.2 Citations to works preparing membranes with self-assembly would be useful, e.g. Vries et al. J. Am. Chem. Soc., 2004, 126 (14), pp 4488–4489.
Section 3.3.3. first line "the system should be minimized (without dynamics, at 0 K)" -> "the system should be energy minimized (without dynamics)" Here and elsewhere it should be specified that the energy is minimized. I think that temperature is not really defined here, so it should not be reported.
Page 5, last paragraph of section 3.3.3 Citation to a good block averaging and autocorrelation analysis protocol would be useful.
Page 6, section 3.4.1 "The region numbering scheme proceeds from the exterior water layer to the bilayer midplane, and the typical density hierarchy is as follows: region 2 ("interface")
Page 6, section 3.4.1 There is -center option in g_density, right?
page 6, section 3.4.2 mdstress.org package is asking to cite also other references than [59].
Page 7, right column, second paragraph Citation to LOOS would be useful.
Page 9, after equation (12) Citation to spontaneous curvature experiments would be useful.
Page 19, left column, last sentence "When simulation and experiment do not agree and one is fairly confident in the simulation procedure and results, the experimental results can be questioned" I think that this is quite bold statement. Experimental results should be questioned only with very good arguments, not just because they disagree with simulations.
Page 11, PREPARATION OF INITIAL CONFIGURATIONS, fourth box Minize energy ...
Page 11, PREPARATION OF INITIAL CONFIGURATIONS, sixth box suggestion: "Dynamics" -> "Equilibration"
Page 11, PREPARATION OF INITIAL CONFIGURATIONS, seventh box suggestion: "confirm thermodynamic averages" -> "confirm equilibration of thermodynamic averages"
Section 4.1.1 It might be useful to mention/discuss here the above mentioned publications comparing the performance of different atomistic resolution force fields against experiments.
"United atom (UA) force fields remove hydrogen atoms for an increase in efficiency (factor of ∼2), and are becoming competitive with AA force fields in accuracy." Someone might say opposite, because the united atom Berger force field was the most accurate for quite many years.
This may also be a good place to mention that there are libraries of models available, like lipidbook (lipidbook.bioch.ox.ac.uk) and www.nmrlipids.fi.
Page 12, right column, second paragraph "Practically speaking, only two sites per lipid are needed (one hydrophilic and one hydrophobic) to capture amphiphilicity and consequently self-assembly into lipid bilayers." I would be very happy to the citation for this.
Page 13, section 4.1.3 "It is worth noting that certain models, particularly CG ones, sometimes experience difficulty in capturing phase transition temperatures and even entire phases (for example, subgel and ripple for Martini)" Based on Pluhackova et al. [ J. Phys. Chem. B 2016, 120, 3888−3903 DOI: 10.1021/acs.jpcb.6b01870 ] it seems that atomistic models have difficulties as well on this.
Page 14, section 4.4.1 discusses only lateral diffusion so "Nuances to diffusion calculations" might more appropriate title.
Page 14, section 4.4.1, first paragraph "The subdiffusive regime has been shown to exist between the ballistic and random (linear MSD) walk regimes, spanning as many as five orders of magnitude in time and timescales reaching many seconds in multicomponent membranes. While this potentially has crucial consequences for dynamical validation of membrane models, its application to a robust protocol for multiscale lipid models is at this point unestablished." I would be happy to see the citation for this.
Table 1, disadvantages of self-assembly. It would be good to mention that self-assembly is really slow compared to the typical simulation timescales.
Page 15, right column, line 8 "0.1 m" is the unit correct?
Various python packages made for the analysis, like MDtraj (http://mdtraj.org/), MDanalysis (https://www.mdanalysis.org/), etc. would be good to mention somewhere.