tag:blogger.com,1999:blog-1283194685237019772.post7395296799131204083..comments2024-03-26T17:11:28.045+02:00Comments on The NMRlipids Project: Does the glycerol backbone structure depend on initial structure?markus.http://www.blogger.com/profile/05973383391755193687noreply@blogger.comBlogger4125tag:blogger.com,1999:blog-1283194685237019772.post-62743558671038430332016-09-01T14:05:20.865+03:002016-09-01T14:05:20.865+03:00If interested, you can do this yourself. The links...If interested, you can do this yourself. The links to datasets in Zenodo are in the beginning of the post: <br />http://dx.doi.org/10.5281/zenodo.13392 http://dx.doi.org/10.5281/zenodo.30904<br /><br />Anyway, by looking at the differences in dihedral angle distributions reported in https://github.com/NMRLipids/lipid_ionINTERACTION/issues/8,<br />it seems quite clear to me that the difference comes from the dihedral.Samuli Ollilahttps://www.blogger.com/profile/06106569992787533569noreply@blogger.comtag:blogger.com,1999:blog-1283194685237019772.post-3734247016148873772016-09-01T12:19:01.867+03:002016-09-01T12:19:01.867+03:00Can you check, what the area per lipid is and how ...Can you check, what the area per lipid is and how it fluctuates in either of the two "inconsistent" simulations? This would be an easy source of the observed discrepancy. Josef Melcrhttps://www.blogger.com/profile/12422047696918877379noreply@blogger.comtag:blogger.com,1999:blog-1283194685237019772.post-63720883435942719672016-05-30T15:34:49.063+03:002016-05-30T15:34:49.063+03:00I am no sure anyone would like to include a "...I am no sure anyone would like to include a "systematic study of conformational sampling of lipids" into the current publication. <br /><br />My personal opinion on this is that it would be the best if we use good initial conditions (meaning close to the observed mean conformation) and add a note of awareness to the discussion -- that we see for example Ulmschneiders model not to equilibrate within xx ns into the proper average conformation and hence the model might easily provide non-ergodic systems. <br /><br />I wouldn't do much about the previous paper until we have at least some numbers to add and make the argument stronger. Josef Melcrhttps://www.blogger.com/profile/12422047696918877379noreply@blogger.comtag:blogger.com,1999:blog-1283194685237019772.post-8236425816361072902016-01-24T00:40:19.336+02:002016-01-24T00:40:19.336+02:00The 2015 paper by Ferreira et al., I think, showed...The <a href="http://dx.doi.org/10.1063/1.4906274" rel="nofollow">2015 paper by Ferreira et al.</a>, I think, showed that PC lipids sample the relevant parts of their conformational ensemble in 50 ns. And, therefore, a correct MD simulation starting from a <b>typical</b> conformational distribution of PC lipids should also sample the relevant conformational ensemble in 50 ns, and thus produce the correct C-H order parameters.<br /><br />However, if the simulation is started from a pathological conformational distribution, I think we are looking at a process that is (or at least could be) very different from the sampling process studied in the NMR experiments of Ferreira et al.<br /><br />That is, when we start from the pathological case, we have an equilibration process from an unlikely part of the conformational space to the typically populated part. This process could (at least in principle) have considerable different time scale than the (50 ns) equilibrium sampling process.<br /><br />Now, it might be of interest to test if (and how fast) the better force fields (such as C36) are able to find the realistic conformational ensemble, when started from a pathological state. However, as we do not know the time scale expected for such a process in real life, these studies would be mostly of interest in terms of knowing how sensible the conformational ensembles observed in MD simulations are on the initial conformations. (But they would not tell us if this returning process happens on a physically correct time scale.)<br /><br />However, what seems to clearly be the case is that we should test the various models starting from initial conformations that are as close to the real life conformations as possible. If the models are able to keep the lipids in this part of the conformational space and sample it properly (give correct OPs), then they should be judged as good force fields.<br /><br />Therefore, I would suggest running simulations of all the badly ranking force fields starting from the final conformations of the C36 simulations.markus.https://www.blogger.com/profile/05973383391755193687noreply@blogger.com