A few problems with MSpace

Hi Judit/Miklos,





We just need to address a few problems with MSpace.





1. Lipid discrimination


We changed the segid of the lipids to Lip1,lip2 etc... it doesnt work.





2. Chain id for lipid molecules


I am not sure why the applet displays chain id as 'chain A' for the DMPC lipids. It has same formats of segid with the other proteins, so the applet


should display it as 'chain DMPC', like the other 'chain PROA', 'chain PROB', and so on. Right now, If the PDB files has more then one chain with segid DMPC, DPPC and POPC, MSpace combines all these segids and displays it at Chain A.





3. Secondary structure without selection panel


I happen to test displaying secondary structure without selection panel and it turned out that the applet couldn't display it. Can this also be fixed?





4. Line Width


When we have a water box of about 10000 water molecules around a protein, it gets difficult to see the protein. Is there a parameter to decrease the line width of molecules to 0.5 from 2? is there also a way to set this in the html coding, without the javascipt like


<param name = "water.lineWidth" Value = '0.5'> ?


Because setting the value in the java script takes a lot of time.





5. Secondary Structure Recognition


Is there a way to recognize the Secondary Structure? As a temporary solution, we used to copy the helix/sheet information from the original pdb. but it doesnt work in the new release. This feature is really important as we will be submiting scientific papers in journals using images taken from mspace.





6. Snap snot.


Is there a feature that allows you to save a snap shot of the molecules from the applet? we must be able to save the snap shot in various formats, and the pictures must be of high quality.





7. Ion


MSpace has problems reading ions. So we changed the pdb entry from ATOM to HETATM. and Mspace reads the ions as Ligand. Also, if a pdb has 100 CL or K ions, MSpace shows up 100 chains in the selection panel. Is it possible to overcome this? all the CL ions must have just one chain and K ions must have another chain.





8. Speed.


When Mspace is opening large PDB files, is it possible for mspace to display the structure faster? I think the problem comes from having too many water molecules. Is there any kind of a trick for mspace to open such files faster?

Hi Vidyashankara,





thank you very much, that's what I call detailed feedback of great value!


I'll try to answer some of the questions, the rest will be answered by Miklos.





3. Secondary structure without selection panel


I'll check and fix this quickly.





4. Line Width


It has to be investigated, because I'm afraid of checking for so many possible parameters in the applet.


Until this, we can change the default linewidth to 0.5 if this is suitable for you.





6. Snap shot


Yes, it is available. I'll do a short tutorial about this and place the link here.


Until that, the short answer is that it can be reached from the main menu:


'Save screen...' creates an image having the same size as the canvas in PNG, JPEG or BMP format. It is a slower method, since it does full-screen antialiasing.


The other option is 'Export Image...' where you can save images in arbitrary resolution, though there are some limitations in functions that will be described in the tutorial.


Please report us if you have any trouble with saving images.





7. Ion


Certainly, this can be done.





8. Speed


To be honest, we haven't tried so many water molecules. We will test it and optimize the visualization code. Can you help us by sending a pdb that opens slowly?





Thanks,


Judit

Hi Vidyashankara,





you guys keep us really busy with mspace! - very good, thank you.





1. Lipids: apologies for the confusion, it was only a question/suggestion from us, but if you think you can adopt the idea we implement it.





2. We'll investigate and fix it. Though a sample PDB would be rather helpful.





5. HELIX/SHEET record copied from original PDB: chain ids are inconsistent, it is A, B etc in the HELIX and SHEET records, while PROA, PROB etc is given in the ATOM/HETATM blocks. We make a quick-and-dirty workaround for that. The development of the proper solution, namely the recognition of the secondary structure is scheduled for the next few weeks, priorities also depend on your requirements.


Btw: when and in which journal you plan to publish?





7. This should be optional though. In most cases ions should be separate entities. We may introduce another level of hierarchy in the selection panel. We'll see, he have not analysed possible solutions deeply.





Best regards,


Miklos

Hi, I am sunhwan and working with Vidya these days.


We almost finish developing core funcionality and waiting testing and validating to finish. What we want to do is seamless integration of command line MD tools with web interface. In that sense, your MarvinSpace Applet is also very important to us as a visualization tool.





Here I attach the pdb file, which is not readable with MarvinSpace without any treatment. Below is the number of treatments to make it readable. If we can read it without any treatment, it would be nice. But as far as we can read it, it is okay.





1. Our output may contains ions


-. first, we have to change ATOM keyword to HETATM and change residue names from POT and CLA to K and CL, respectively for ion.


-. Could you check CL atoms are readable in the latest mspace applet? For me, potassium is okay, but whenever PDB has chlorine, it can't read.


-. We want mspace recognise same kind of ions as a chain. Our system will have 40~60 ions, and mspace displays each of them in selection panel. It's hard to click 'secondary structure' or 'tip3' in selection panel due to ions.





2. Reduce number of water molecules


-. To make it faster, we usually reduce number of water molecules to 1/10 of its original size. But it is not pretty.


-. So the speed issue arises here.





3. Secondary structure


-. It is the main issue for us now. Because we are going to publish, but we can not make a snapshot with tiny little string-like secondary structure.


-. Now I understand what had happened for our trick. (SEQRES/HELIX)


-. If you can develop secondary structure recognition, it is the best. But if it will takes time, do you have any idea to bypass the problem?





4. Lipid recognition


-. Lipid id problem is mentioned above.


-. What we really want to do is, changing display options such as line width for lipids. (except protein) If you open the attached pdb, protein is surrounded by lipids. To impress the user where is the protein, we need to make the lipid molecules thinner.








Thank you.

Hi Sunhwan/Vidyashankara,





you had some problem with secondary structures and selection panel (3.point).


With a quick test I was unable to reproduce it. Can you help me with more details? Maybe this problem is more specific.





We had a test relase that can be downloaded from http://www.chemaxon.com/test/marvin.





Thanks,


Judit

Hi, Thanks,


Secondary structure without selection panel is working fine.


I don't know, maybe I did something wrong.





And could you take a look the attached pdb?


I don't know why mspace complaining CL ion.


Potassium and Zn is okay, but if I put CL, it refuses to read.





Thanks,

Hi,





Just one small change in hte electrostatic potential feature.





Right now, a positive charge shows red and a negative charge shows blue. Can we reverse that?





Thanks


Vidyashankara

Hi Sunhwan and Vidyashankara,





regarding Sunhwan's recent list of requests:





1. We will try to resolve all issues regarding groups of ions and element symbols.





We are not sure on what basis should CLA be recognized as CL? Can CLB also occur in such PDB-like file, and that is also CL? Is - for instance - MGA also a valid element/ion symbol? Can you generalise this somehow in terms of simple rules?


Do you use other synonyms for element names? E.g. Sod (ie sodium) for Na?








2. We will need to change the internal representation of water molecules. Currently we use the Marvin Molecule object, which cannot handle large number of atoms efficiently - because it was developed to store small molecules of less than 50 atoms on average.


This is a fairly complex task and it will take some time to implement it.








3. In order to privide a fast, though temporarily solution, we will make a quick workaround that will allow you to use the original helix/sheet records from the pdb file without the need to modify them. This should support your publication needs, though with some manual intervention.


We understand that this is not sufficient for the production system you are developing, thus


the secondary structure recognition will be implemented (for proteins, helices and sheets only, we may consider turns later). However, we cannot deliver it sooner than end of next week - even this time is flexible.


(Both Judit and myself are involved in an other project in which we have strict deadlines....)





4. Regarding lipids, the recognition of LIP will be implemented soon and the lipid part of the macromolecule will independently be accessible in the selection panel.


In the longer term, we may need to implement a better (more efficient) internal representation to store lipids (similarly to the way how nucleic acid and amino acid polymers are stored).








Summarising: all requests are approved, all will be implemented. None of them requires too much work itself, but all put together takes substantial amount of time, and as said above other commitments prevent us from delivering the solutions within days... We'll try our best and we will inform you about the status of the development.





Regards,


Miklos, Judit

The color palette can be changed by

Code:

MSpaceApplet.setProperty("Surface.Palette", "Red to Blue");

The following ones values can be used:


"Red to Blue", "Blue to Red",


"Rainbow", "Reverse Rainbow",


"Blue to Green", "Green to Blue",


"Fire",


"Blue",


"Green",


"Yellow"





This setting affects the grid visualization as well.

About setting the line width without the javascript...what do you think about the following:





MarvinSpace has its own scriping capability, which we use for example in our Java Web Start demos at http://www.chemaxon.com/marvinspace/html/webstart.html.





The point is, that every setting is identified by strings, e.g. "Surface.Palette" or "Water.LineWidth", etc.


These settings can be collected in a text file, and MarvinSpace can execute it. It is able to write it too, so you don't have to write it by hand though it is not difficult.


These scripts can easily be used in multiple applets, and it is easier to maintain.


An example applet can be found here: http://chemaxon.com/marvinspace/applet/chapplet.html


It uses the following log file: http://chemaxon.com/marvinspace/applet/ch.txt





If you consider it useful, we can help you in the details.


Judit

Hi,


Sorry for late response. We were busy preparing the first presentation of our website.


We appreciate your quick and superb efforts. That is why we strongly want to stick with marvin space.

mvargyas wrote:

1. We will try to resolve all issues regarding groups of ions and element symbols. We are not sure on what basis should CLA be recognized as CL? Can CLB also occur in such PDB-like file, and that is also CL? Is - for instance - MGA also a valid element/ion symbol? Can you generalise this somehow in terms of simple rules? Do you use other synonyms for element names? E.g. Sod (ie sodium) for Na?

We are working with CHARMM, a molecular dynamics tool. Ion names like CLA and SOD are come from CHARMM parameter file. So far, it has 11 ion parameters. They are SOD (sodium), POT (potassium), CLA (chloride), CAL (calcium), MG (magnessium), CES (cessium), ZN (Zinc), FE (Iron), HE (Helium), NE (Neon), and CLAL(Chlorine).





Thanks


Sunhwan

Hi Sunhwan,





we've been also busy in another project... We are glad that you want to stick to MarvinSpace.


In the next few weeks we can devote limited resources to the MarvinSpace project, but your requests have the highest priority on our to-do list.





Thank you for the clarification regarding the CHARMM specific atom/ion type names.





Regards,


Miklos

Hi Vidyashankara, Sunhwan,





we have implemented secondary structure recognition based on the Ramachandran plot.


It will be included in the next release, until then a trial applet is available here.





Please let us know if this is suitable for you.





Btw. did you succeed in taking screenshots?





Best regards,


Judit

Thanks, but the marvin space on the url is not working. Could you please check again?





I've tested with my Mac OS, Windows, and Linux.


All I can see is the message "Marvin Space applet is started" in the status bar. Nothing just showed up.





Thanks,


Sunhwan

Judit wrote:

Hi Vidyashankara, Sunhwan, we have implemented secondary structure recognition based on the Ramachandran plot. It will be included in the next release, until then a trial applet is available here. Please let us know if this is suitable for you. Btw. did you succeed in taking screenshots? Best regards, Judit

Hi Judit,





Sorry for such a late response. I just got back from a vacation to India. I hope you had a great christmas and new years.





The screenshot feature works great, atleast when i tried it a month ago. So we shouldnt have any problem with that.





We will start to test the new secondary structure recognition system intensively from monday and i hope to give you back a positive feedback by tuesday/wednesday.





As sunhwan pointed out, the applet loads but nothing shows up. I am attaching a screenshot to make your work easier. I am running Firefox 2.0.0.1 on Windows Vista. I trust sunhwan as tried it on Mac OSX, Windows XP and Linux.





Hope to hear from you soon!


-Vidyashankara

Hello,





indeed, the applet is not working for me either. Apologies, we'll fix that soon.





Regards,


Miklos

Hello Vidyashankara, Sunhwan,





good to hear from you, hope you had a great journey in India!





Sorry about the problem, the jogl version we currently use is not working on windows with java 1.5_10, so I'm using a newer jogl for a while to test.


I've uploaded the newer jogl.jar by chance without any native libraries.





I've placed the older jogl.jar there so it should work now, except with the specific java on windows.





The secondary structure detection based on Ramachandran plot is not 100% accurate, but it can be easily modified if need be.





Thank you,


Judit

Thanks for your correction.





It works fine now. Here are few things that I found.





1. Additional lines?


When I load pdb file, sometime I can find additional lines which you can find in attached image. (filename: picture 1.png, picture 2.png, Picture 6.png)





2. Secondary structure for 1uyn


I've tested couple of pdb files that I have now, but it couldn't draw secondary structure for 1uyn. I'll attach pdb file and image files from marvinspace and rasmol. (1uyn.pdb, picture 3.png, picture 4.png)





3. Closing components


This may not be relevant to secondary structure representation. When I load a big pdb and open another file, and selecting close components, marvinspace hangs up. Is this a problem of closing component? I will attach the pdb file that I had problem. (2a65.pdb, http://www.charmm-gui.org/2a65.pdb)





4. Water


This is also not relevant to secondary structure. But since we are dealing with membrane protein complexes, huge amount of water molecules are inevitable. We will be glad if you can improve water displaying capacity. Now we are eliminating water molecules from pdb which will be used for marvinspaces, but it will be better to have water.





Appreciate your kind efforts,





Thanks,


Sunhwan

Thank you for the fast feedback,





1. we will fix the additional lines bug asap.





2. indeed, there was a problem in the algorithm. Thank you very much, it is corrected.





3. If I understood well, I should open 2a65.pdb, and open any other file with selecting 'close components' in the appearing dialog.


Please correct me if some more steps are necessary, as I couldn't reproduce the problem this way. (maybe it was due to the previous problem too)





4. Water: this is the next thing we will work on.





Judit

Thanks your for your quick response.





2. resolved.





3. resolved.


It is considerably slow in my mac os, but at least it works.


and it is pretty fast on linux. I was impressed.


but, strangely, my windows machine couldn't able to load the applet. (sorry, but can you check with windows again?)





Thanks again.


Sunhwan

It should work now.





Water speed-up is in progress, I attach a test file containing >15K water molecules.


The trial applet loads it very fast, though I had to increase the JVM memory (-Xmx512M) in the Java Control Panel to let it work.


I'd like to ask whether you can test it with a protein in a water box you mentioned earlier.





Thank you in advance,


Judit

Hi Judit,


We are in the process of writing the manuscript for Charmm-Gui and it should be done by the month-end. We will be including images taken by marvinspace. I would like to thank you for all the support you have given us. Also, We would really appreciate it if you can make the latest applet, with all the fixes and secondary str recognition, for download so we can add it to the website?


Thanks again


Vidyashankara & Sunhwan.

Hi Vidyashankara, Sunhwan,





We're glad we could work together, I hope we can continue it in the future.


The latest updates will be included in Marvin 4.1.6 which will be released next week (it's already in test phase).


Would it be possible to see the results after you finish your work?


Thank you,


Judit

Hi Judit,


We are still in the process of writing our manuscript. I will ask our research advisor and get back to you about disclosing test case database. In the meantime, we would like to setup the site to start working with the new Mspace. if i am not wrong, its supposed to be out sometime this week rt?


Thanks


Vidyashankara

Hi Vidyashankara,


Yes, it's ready soon, I'll post here as soon as the release is out.


Judit

Hi Vidyashankara,





I'm very sorry that the release is late, can you perhaps use the test version until it is ready?


It can be found here: http://www.chemaxon.com/test/marvin/download.html.





Judit