[gmx-users] phi / psi values using g_chi

[chris.neale at utoronto.ca]

I was either using using 3.3 or 3.3.1, certainly not 3.1.4 or 3.2.1.

It looks like the change was in 3.3.1, possibly by Erik:

[tera:~]?diff gromacs-3.3/src/tools/gmx_chi.c  
gromacs-3.3.1/src/tools/gmx_chi.c
2c2
<  * $Id: gmx_chi.c,v 1.3 2005/02/16 17:39:27 hess Exp $
---
>  * $Id: gmx_chi.c,v 1.3.2.2 2006/04/02 15:09:37 lindahl Exp $
148,172c148,170
<   for(i=0; (i<nl); i++) { /* Phi */
<     dl[i].j0[edPhi] = n/4;
<     if (dl[i].atm.H == -1)
<       id[n++]=dl[i].atm.minC;
<     else
<       id[n++]=dl[i].atm.H;
<     id[n++]=dl[i].atm.N;
<     id[n++]=dl[i].atm.Cn[1];
<     id[n++]=dl[i].atm.C;
<   }
<   for(i=0; (i<nl); i++) { /* Psi */
<     dl[i].j0[edPsi] = n/4;
<     id[n++]=dl[i].atm.N;
<     id[n++]=dl[i].atm.Cn[1];
<     id[n++]=dl[i].atm.C;
<     id[n++]=dl[i].atm.O;
<   }
<   for(i=0; (i<nl); i++) { /* Omega */
<     if (has_dihedral(edOmega,&(dl[i]))) {
<       dl[i].j0[edOmega] = n/4;
<       id[n++]=dl[i].atm.minO;
<       id[n++]=dl[i].atm.minC;
<       id[n++]=dl[i].atm.N;
<       id[n++]=dl[i].atm.Cn[1];
<     }
---
>   for(i=0; (i<nl); i++)   {
>         /* Phi, fake the first one */
>         dl[i].j0[edPhi] = n/4;
>         if(dl[i].atm.minC >= 0)
>                 id[n++]=dl[i].atm.minC;
>         else
>                 id[n++]=dl[i].atm.H;
>         id[n++]=dl[i].atm.N;
>         id[n++]=dl[i].atm.Cn[1];
>         id[n++]=dl[i].atm.C;
>   }
>   for(i=0; (i<nl); i++)   {         /* Psi, fake the last one */
>         dl[i].j0[edPsi] = n/4;
>         id[n++]=dl[i].atm.N;
>         id[n++]=dl[i].atm.Cn[1];
>         id[n++]=dl[i].atm.C;
>         if ( i< (nl-1) )
>                 id[n++]=dl[i+1].atm.N;
>         else
>                 id[n++]=dl[i].atm.O;
...
<snip>

Note that I am not sure if 3.3.1 will give you what you expect for the  
terminal residues, even if they are capped. I suspect that it will not.

In the link that I mentioned earlier, I discuss "...plot the dih.rest. energy
output from g_energy and scatter plot this with the actual dihedral value from
g_chi for a particular umbrella". If there is scatter from the expected energy
function then you may suspect that g_chi is still incorrect.

I switched to g_rama for my work and never looked back, so don't take  
my word for it that g_chi is correct now, I just took a really quick  
look.

Chris.

-- original message --

Hi Chris,

Thanks for the link.

I then do not understand what is happening because I get the same
exact values
with g_chi and g_rama! Are you sure you were using the gmx-3.3.1? And
not an
earlier version?

Note that in the code of gmx-3.1.4 and gmx-3.2.1 the non-standard
definition is used
but in the gmx-3.3.1 the code is changed and the standard definition
is used.

Anyone would have changed the code and forgot to update the comment on
the
choice of the atoms to calculate the phi/psi angles in g_chi ?

XAvier.
---------------------
Data:
g_chi / from the phi-psi files for Ala15
10000 -112.924 42.8174
10100 -114.15 57.1768
10200 -82.0442 -11.6453
10300 -99.8691 36.5998
10400 -83.1319 -20.2397
g_rama / extracting the Ala15 data
10000 -112.924  42.8174  ALA-15
10100 -114.15  57.1768  ALA-15
10200 -82.0442  -11.6453  ALA-15
10300 -99.8691  36.5998  ALA-15
10400 -83.1319  -20.2397  ALA-15
10500 -104.275  12.6959  ALA-15