[gmx-users] Self made GDP topology file....deforms during EM ...and the reason

[Zhenting Gao]

Hi  gmx-users,                                     

     Sorry, I have sent several mails with unreadable chraracters to the maillist due to my email client software.   
     And thanks for the kind replies.   

     Luckily, after days of struggle, I find the reason of the deformation. Hereafter I repeat the problem first.   

     A GDP is in my protein complex.   
     I choose Gromos 96FF to do MD, and made my own GDP topology file in Gromos 96 FF.   
     First, I get the topology from PRODRG server.        
     Then I use Gaussian to recalculate the atom charge.        
     At last I refer to the topology of ATP and GUA in ffG43a1.rtp to modify the topology file for GDP.   
     To validate the topology file for GDP, I included the GDP in a water box and performed a 20 ps EM and heating.   
     However, the GDP deforms severely during the EM, and even crashes if I run a heating process.(it deforms whether I add ions or not.)   
     In contrast, the topology files generated by pdb2gmx for ATP, GUA and DGUA make the structure ramain good during a 300K heating process.      
     Why it deforms? It is the wrong force parameter? Or the wrong atom charge?    

     I did a comparison of the definition of atom, bonds, angles, dihedrals between my GDP toplogy and the GUA topology in the ffG43a1.rtp.   
     And I find out that   
          1.every thing is the same(almost) except the difference of atom charges   
          2.charge difference of atoms that deform the structure(C2, N2, N3, N1) is very large(C2 0.715, N2 0.354, N3 0.462, N1 0.511).   
     Furthermore, I modified the GDP atom charge acoording to the GUA and ATP and did a 2ns heating(300K), the structure remain intact.   
     So I believe it is the charge that caused the deformation.   

     I searched this maillist, and find some similar questions about the charge calculation.   
     http://www.gromacs.org/pipermail/gmx-users/2003-August/006691.html   
     http://www.gromacs.org/pipermail/gmx-users/2003-September/006970.html    
         
    It seems that QM(e.g. 'CHELPG' or 'Merz-Kollman' methods in Gaussian) charges are too big in some cases.Though the 'big' charge may not result in a deformation( my collegues use QM charges well), they are not appropriate in the cases of ATP, ADP and GDP.( My collegues found before that self made topology for ADP and ATP with QM charges also defrom. 
http://www.gromacs.org/pipermail/gmx-users/2004-March/009747.html).  

     Everyone should know that the atom charge is really important for an MD simulation; the QM charge may not be the best choice. 
     And I really expect that someone can give details in how to calculate the atom charge for gromacs topology. 

     Best wishes.  

Yours sincerely,                                      
Zhenting Gao                                      
zhentg at 163.com                                       
2004-11-9                                

------------------------------------                                      
Drug Discovery and Design Center,                                      
Shanghai Institute of Materia Medica,                                      
Chinese Academy of Science                                      
P.R. China                                    
http://www.dddc.ac.cn/group/zhentg.htm