Tuesday, December 22, 2015

Towards submission of NMRlipids II publication (lipid-ion interactions)

The manuscript from NMRlipids II project "The electrometer concept and binding of cations to phospholipid bilayers" is (in my opinion) quite close to the submittable version. There are still some details to be fixed which are listed in the ToDo list in the manuscript.

All kind of comments are welcomed also from people who are not contributors yet. My goal is to get manuscript submitted January 2016 so this is almost the last change to present major comments. Manuscript will be submitted soon, all modifications and comments should be done before 19.2.2016.  Manuscript will be submitted soon. As in all the subprojects, the files are available in the project GitHub repository and contributions can be done by commenting this blog or using GitHub. Especially, if you are doing language or other text editing the most convenient way is to directly modify the text file and make a pull request.

My current idea is to submit the manuscript first to the Chemical Science which is a high impact journal by the Royal Society of Chemistry. By doing this we would find out their policy towards open collaboration publications which was also the motivation to submit the NMRlipids I publication to a American Chemical Society journal, see discussion in Towards first submission to journal post. All comments related to this are also welcomed.

Monday, September 28, 2015

NMRLipids V project: Review about validations of membrane MD simulations

I have been invited by Tomasz Rog and Ilpo Vattulainen to write a review article to a special issue of BBA-Biomembranes about validations of membrane MD simulations. Since the topic is very closely related to the NMRLipids project I asked a permission to use the open collaboration method, introduced in this project, to write also the review. The guest editors and BBA staff supported my suggestion. I was also happy to note that the Elsevier policy contains the sentence "Authors can share their preprint anywhere at any time."

Thus, the writing of this review is now the NMRLipids V project. The deadline for submission is 14th of November 2015.

The focus of the review will be the best data (Order parameters and Spin lattice relaxation from NMR, form factor from scattering) and methods to validate the atomistic resolution structure and dynamics in membrane MD simulations. The focus is not in the details of different models or in actual comparison between force fields (these are discussed in other NMRLipids projects). Also, only structure and dynamics will be covered, not thermodynamics (phase transition temperature, calorimethric data, etc.).

The first draft of the manuscript with (very preliminary) introduction is already in the project GitHub repository. It is straighforward to me to write the NMR part based largely on the content of this project, and also on our recent work. However, I am requesting help especially for the scattering part. There are number of things which are not perfectly clear for me. Many of these issues have been already came up in the discussions related to the NMRLipids III project, thus writing the review will strongly support that project as well.

The recently introduced workflow will be used in this project, i.e. the up-to-date manuscript is all the time in GitHub containing pointers to the more specific issues and related discussion. The best overview of the project status should be obtained by reading the manuscript and the linked issues.

The original rules for the authorship will be applied to the people who comment the blog post containing NMRLipids V in the title or contribute (commit or comment) to the NMRLipids V GitHub repository.

Thus, everyone who has contributed to the project by communicating via this blog or GitHub will be offered coauthorship. The final decision on authorship will be to the invited individuals themselves. They should base their decision on self-assessment of their scientific contribution to the project. The order of all names in the final manuscript will be alphabetical. O.H. Samuli Ollila will be the corresponding author.

Monday, July 20, 2015

Revision requested for the first manuscript

The Journal of Physical Chemistry has sent us the reviewer comments on the first manuscript written based on the work in this blog. Major revision is requested before potential publication, however, based on the comments I am quite optimistic for the acceptance. I have asked the editor, if we can make the reviewer comments publicly available, but am yet to receive a reply. In the meanwhile, I wrote preliminary answers to the reviewer comments (available on GitHub) and sent the reviewer comments to the authors of the publication. Other followers can try to guess the original reviewer comments from the replies, until we hopefully get a permission to make also the original comments available.

If you want to comment or modify the answers, you can comment this post or (preferably) modify  directly the answer on GitHub and make a pull request.

In addition to replying the comments, there are some other things to do before the submission of the revised version:
  • Make the table of content figure. I do not have any good idea for this and also I do not have much time right now, so any ideas and actions are welcome.
  • Clean the GitHub repo a bit and create a doi for it.

Wednesday, March 25, 2015

Mapping scheme for lipid atom names for universal analysis scripts

During project we have generated unique collection of simulation data in the Zenodo community which can be openly used for different kind analyses. Analysis of this data is straighforward for expert, however generating scripts may be tedious since the atom naming convention is different between different models. To ease this I have generated a atom name mapping convention which should allow the usage of the same analysis script for different force fields with minimum effort. This has been done bearing in mind the analysis we have to do for the current activities in this blog, but also the wider usage beyond this project. In this post I describe the idea of the mapping and show some examples where I have used it. I hope that other people try to use this as well and tell if there are suggestions for improvement.

The mapping consist of a file which has the universal atom names in the first column and force field specific atom names in the second column. Here is example from the beginning of the mapping file for the CHARMM36 force field:

Universal name                   force field name
M_G1_M                                     C3
M_G1H1_M                               HX
M_G1H2_M                               HY
M_G1O1_M                               O31
M_G1C2_M                               C31
M_G1C2O1_M                             O32

Universal atomnames always starts with "M_" flag and ends with "_M" flag to ensure that the scripts using grep command never confuses between them and force field specific names. In the actual naming convention between the flags, the first two characters define in which glycerol backbone chain the atoms attached (G1, G2 or G3). Then the naming goes like this:

M_G1_M               ;glycerol backbone C1
M_G1H1_M         ;hydrogen attached to glycerol backbone C1
M_G1H2_M         ;hydrogen attached to glycerol backbone C1
M_G1O1_M         ;first atom (oxygen) in sn-1 chain
M_G1C2_M         ;second atom (carbon) in sn-1 chain
M_G1C2O1_M    ;oxygen attached to second atom (carbon) in sn-1 chain
M_G1C3_M         ;third atom (carbon)  in sn-1 chain
M_G1C3H1_M    ;first hydrogen attached to third atom (carbon)  in sn-1 chain
M_G1C3H2_M    ;second hydrogen attached to third atom (carbon)  in sn-1 chain

So the third character tells the atom type and fourth character tells the counting number from the glycerol backbone carbon. If there are hydrogens or other atoms attached to the main chain, those will be added to the end of the naming, for example, in these cases:

M_G1C2O1_M    ;oxygen attached to second atom (carbon) in sn-1 chain
M_G1C3H1_M    ;first hydrogen attached to third atom (carbon)  in sn-1 chain

The G3 atoms are the headgroup atoms, thus they are little bit more complicated. Examples of complete mapping files for CHARMM36 and MacRog models can be found from GitHub.

I have used this mapping for couple of cases now and I think that it works.
Below is example script which calculates the order parameters for acyl chains for CHARMM36 model in low hydrated conditions. The main point is that only changes required to use the script for different systems are the file names for simulation files, output files, mapping file and the order parameter analysis script location. These are 7 lines after "#Define file names" comment. The gro_OP.awk and mapping files can be found from GitHub. It is tedius to generate the mapping files, however, once done the analysis of different properties becomes very straightforward.

Example script:

#Dowload files from Zenodo
wget https://zenodo.org/record/13945/files/popcRUN4.tpr  
wget https://zenodo.org/record/13945/files/popcRUN4.trr
#Define file names
#Make gro file which can be used to calculate the order parameters
echo System | /home/ollilas1/gromacs/gromacs465/bin/trjconv -f $trajname -s $tprname -o $trajgroname -pbc res -b 0
#This is loop over sn-1 carbon segments
for((  j = 3 ;  j <= 16;  j=j+1  ))
    #This greps the force field specific atom names using the mapping file
    Cname=$(grep M_G1C"$j"_M $mappingFILE | awk '{printf "%5s\n",$2}')
    H1name=$(grep M_G1C"$j"H1_M $mappingFILE | awk '{printf "%5s\n",$2}')
    H2name=$(grep M_G1C"$j"H2_M $mappingFILE | awk '{printf "%5s\n",$2}')
    #Calculate order parameters
    H1op=$(awk -v Cname="$Cname" -v Hname="$H1name" -f $analFILE $trajgroname)
    H2op=$(awk -v Cname="$Cname" -v Hname="$H2name" -f $analFILE $trajgroname)
    #Print results to file
    echo $j $H1op $H2op >> $sn1outname
#This is loop over sn-2 carbon segments
for((  j = 3 ;  j <= 18;  j=j+1  ))
    #This greps the force field specific atom names using the mapping file
    Cname=$(grep M_G2C"$j"_M $mappingFILE | awk '{printf "%5s\n",$2}')
    H1name=$(grep M_G2C"$j"H1_M $mappingFILE | awk '{printf "%5s\n",$2}')
    H2name=$(grep M_G2C"$j"H2_M $mappingFILE | awk '{printf "%5s\n",$2}')
    #Calculate order parameters
    H1op=$(awk -v Cname="$Cname" -v Hname="$H1name" -f $analFILE $trajgroname)
    H2op=$(awk -v Cname="$Cname" -v Hname="$H2name" -f $analFILE $trajgroname)
   #Print results to file
    echo $j $H1op $H2op >> $sn2outname

The script and output files (OrderParamSN1lowHYD.datOrderParamSN2lowHYD.dat ) can be found also from github.

The results from the above script together with the full hydration results are also shown in Fig. 1

Fig 1. Order parameters for acyl chains calculated from trajectories availabe in Zenodo collection using the mapping file and similar scripts as above. Also experimental results by Ferreira et al. for full hydration are shown. Dehydration induced ordering of acyl chains qualitatively agrees with experiments for DMPC [Dvinskikh et al., Mallikarjunaiah et al.].

I have also ran similar script to calculate acyl chain order parameters from different simulation data with different cholesterol concentrations available in the Zenodo collection. These scripts and produced results are available in GitHub (CHARMM36, MacRog). The results are summarized in Fig. 2
Fig 2. Order parameters for acyl chains calculated from some trajectories with cholesterol availabe in the Zenodo collection using the mapping file and similar scripts as above. The experimental data is from Ferreira et al. For more discussion about CHARMM36 and MacRog in GitHub.

This is quite technical issue, and I am not sure if this presentation is understandable, so please do not hesitate ask further clarifications.

Tuesday, March 17, 2015

Towards first submission to journal (2)

I have now added the new version of the "Towards atomistic resolution structure of phosphatidylcholine glycerol backbone and choline headgroup at different ambient conditions" manuscript in the GitHub:

The filenames for this version are HGmodel_ACStemplate.tex and
HGmodel_ACStemplate.pdf. The latter can be found also from the Dropbox link.

The manuscript has been now put in the ACS template and the current idea is to submit it first to the Journal of American Chemical Society (JACS), for the reasons discussed in the Towards first submission to journal post. I have now modified the abstract and conclusions to be more suitable for this journal. The journal for the first submission can be still discussed if there are objections from the authors.

Otherwise, I hope that the current version will become the final version for the first submission once the issues written with red in the manuscript are fixed.

If you want to proofread, or otherwise comment the manuscript, the most convenient way is that you modify directly the HGmodel_ACStemplate.tex file and make a pull request to update the changes also to the GitHub. More traditional options for commenting/modifying can be used, however the changes should be made somehow visible.

We have also started to use the issue tracker in GitHub and it seems to be quite convenient, at least from my point of view:

There are issues related to this version as well, and do not be shy to add more. I think that especially more detailed and technical issues are convenient to deal with the issue tracker.

I am now also writing the draft of the cover letter, and I will add this to the GitHub also in the near future.

Monday, March 9, 2015

Current and future activity

The project has expanded, and many things are going on. Here I go through the status of currently active topics, and present some new ideas on which to focus in the near future.

Two publications are currently being written based on results discussed in this blog:

One of the original goals of the project is still in progress:
  • Finding an atomistic (preferably united-atom) force field that correctly describes the glycerol backbone and choline headgroup structure in different conditions. This line of research was discussed mainly by me and Antti Lamberg until July 2014. After that, I have focused on writing the manuscripts on results already obtained. During this time I have learned some relevant details about the (dihedral angle distributions of the) structures, see especially Figs. 5 and 6, and related discussion, in the publication 1. Also, a new united atom model by Tj√∂rnhammar and Edholm, which seems to provide a better structure to start with, was published. I think that this line of research should be continued using the Tj√∂rnhammar14 model as a starting point.

There are several possibilities for the future directions of this project, and I believe that there is a huge potential also on a longer time span. The project will run for at least one more year from now. Further continuation of myself and Markus Miettinen as editors, as well as other long term future directions, depend on the success of our grant applications. For now, I will focus on these two topics that go beyond the original goals of the project:
  • Quantitative quality of lipid–cholesterol interactions in simulations based on x-ray scattering and NMR data. Lipid–cholesterol interactions have been widely studied with MD simulations in the recent years. Qualitatively, simulations reproduce the condensing effect, but it is not clear how accurate, quantitatively, the models are. This is crucial information to judge the credibility of the various simulation predictions. Directly comparing the simulations with different models (made available in Zenodo in this project) against x-ray scattering and NMR results, we can get a pretty clear picture of the quantitative quality of the models. The discussion on this is already going on in the blog with Peter Heftberger and Georg Pabst. I will soon write a blog post discussing this project in more detail.
  • Glycerol backbone and headgroup order parameters for other than phoshatidylcholine lipids (e.g., phophatidylethanolamine (PE), phosphatidylglycerol (PG), phosphatidylserine (PS), glycolipids etc.). I have privately discussed the existence of order parameter data for these and other headgroups at many occasions. The bottom line is that a lot of published data exists, however, they seem to be quite poorly known and difficult to find. To my knowledge, these data have not been reviewed or collected anywhere (not even by Derek Marsh). I think we should, at minimum, collect the available data sets in an easily accessible format. This, with potentially some comparison to simulations, would teach us interesting lessons about the structural differences between headgroups, allowing us to expand on what has been discussed in the literature so far.

In addition, several issues have been shortly touched in the discussions of the blog, but are beyond both the original and the extended scopes of the project. I would be more than happy too see someone to progress, for example, these topics:
  • Temperature dependence of the glycerol backbone and choline order parameters. For example, Andrea Catte has reported interesting data on this and they have been discussed a bit already. There is also some temperature dependence in the data delivered by Fernando Favela.
  • Peter Heftberger shared a collection of x-ray scattering data for different systems. In addition to the lipid–cholesterol interactions mentioned above, there may be many other interesting issues to study with moderate effort by combining the simulation data now available in Zenodo with the scattering data shared in the blog.

Friday, February 6, 2015

The first draft of the ion-lipid interaction manuscript

[update 14.12.2015] Different force field was used for CHARMM36 ions than I tought.

[update 16.4.2015] I have created an own GitHub Repository for this project. The links are updated.

I have now written the first draft of the manuscript considering the ion-lipid interactions discussed in this blog. All the manuscript files (written with LaTex) are in GitHub. The pdf version the manuscript with ToDo list is also available through link https://www.dropbox.com/s/ebiwdsoj2otg4hp/LIPIDionINTERACT.pdf?dl=0, in addition to GitHub.

I think that there are few simulations more which we definitely have to run for this manuscript:
  • Orange model with higher concentration of NaCl (I will do this).
  • CHARMM36 model with CaCl\(_2\) (the data contribution from someone already having this data would be excellent. I if this does not happen, I will do it).
In addition, it would be very interesting to get some results with different NaCl and CaCl\(_2\) concentrations from the modified CHARMM36 ions by Venable et al. According to the paper, the parameters improve the ion binding to the charged lipid bilayers. I think that the data with these parameters would be highly relevant, however I do not have time to do it now.  The force field by Venable et al. is already used in the current CHARMM36 simulations with NaCl, see comment on 14.12.2015.

All kinds of comments about the manuscript are welcomed from everybody (also from people who are not contributors yet). You can submit your comments by commenting this post, or you can directly fork the GitHub repository and make a pull request.

The authorship of the manuscript will be offered for everyone who has commented the blog, according to the "on the credits post". The final decision on authorship will be made by invited individuals themselves based on their self-assessment of their scientific contribution to the project. The separate authorship query is sent to to all contributors with email and authorlist will be updated according to the replies. 

Friday, January 16, 2015

Towards first submission to journal

The new version of the first manuscript is now updated to arXiv: http://arxiv.org/abs/1309.2131v2. I think that the core content of the manuscript is pretty much in its final format already. The current up-to-date version with updated To-Do list can be now found here:

The main issues before submission to a journal are the figures. There have been some discussion that several of them should be clarified, and maybe add some extra figures for clarification as well (Figs. 2, 3 and 6). Also adding some discussion about the results from the force fields which are not among the best three should be considered (ToDo point 20).  The rest of the open issues in ToDo list requires less work. Also proofreading and comments about language would be welcomed.

Anyway, I think that the manuscript is in a state that we should start to consider to which journal we are going to offer it since this slightly affects some writing choices and the style. As always, there are various types of options supported by different arguments. However, due to non-traditional approach used here to produce the results and manuscript, there are more issues to consider than usually.

I have listed here some options and my thoughts about those. Please, express your opinions and suggestions by commenting this post. Ideally, we would reach the consensus. However, if this does not happen, then we will organize a voting.

Different types of options I have thought:

  1. Traditional journal for publications with general interest, e.g. PNAS, JACS, Chemical Science
  2. Traditional journal for publications with general interest in the field, e.g. J. Phys. Chem B, J. Chem. Phys, Phys. Chem. Chem. Phys.
  3. Traditional journal for publications with interest in computational field, e.g. J. Chem. Theor. Comput., J. Comp. Chem.
  4. Modern "open science" journal, e.g. PLOS, Royal Society Open Science.
  5. Historical journal, Philosophical Transaction  

 I have intentionally left out the Elsevier journals since some authors might boycott them.

In addition to the regular issues to be considered when choosing a journal, we have consider also these:

  • Some journals might not accept our manuscript since the content has been already  published (everything is in this blog and in arXiv). To my knowledge this is the first time when this type of work is submitted to a journal (Polymath community seems to publish only in arXiv). Correction: Polymath community has also published in peer reviewed journals.
  • I do not have any money to pay the publication costs. I have not checked the details, but I recall that especially open access journals (e.g. PLOS) is quite expensive, while ACS journals are cheap or even free. The free journals to publish are priority, unless someone wants to pay these costs.
About the regular issues the possible length limitation or limited amount of citations are probably the most important for us (currently we have roughly 12 pages and 152 citations).

I would prefer the first submission to the type 1 journal in the list above, i.e.
journal with publications having general interest. The reason is that I would like to find out if they could publish this kind of work, in principle. This is important question since this might be a barrier for some people to start this kind of collaboration. If the journal rejects our work because it "has been published" already, we know that they are not ready for this yet. If they reject it for some other reason or accept, we know that using the open blog and arXiv publications to generate the publication is not a problem for them. This might encourage people for similar approaches that we are using.