Re: AMBER: Need help... High energies for complex...

From: Lachele Foley (Lists) <"Lachele>
Date: Tue, 9 Sep 2008 10:09:06 -0400

Waqas:

The task you have set out for yourself is complex. It is not a simple
task even for someone who is well familiar with Amber. You need to
carefully examine the results of each step you take. The high
energies you are getting are almost certainly due to an imrproper
setup and not due to some fundamental problem with amber or the force
fields. But, given the complexity of the task, it is nearly
impossible for anyone to guess where the error might be based on a
relatively brief description of your procedure. It could be as simple
as a mis-named atom type or a command issued at the wrong time. Or,
it could be as complex as a complete mismatch between the pdb file and
what leap expects from the various residues.

I recommend, at a minumum, a careful visual inspection of your
structure, using xleap and not an external program like vmd, at each
step in the process of generating your prmtop and coord files. If
doing that doesn't show you the problem, take a close look at one of
your high-energy structures. Are any of the atoms too close or too
far from each other? Are any of the sequences in the sugar or protein
tangled? If that doesn't help, try generating a very small version of
your system. For example, generate a small, 3-5 peptide sequence to
which you can attach a sugar. Then, attach a small sugar to that. Do
you get similar energies? If so, then you have narrowed down the
source of your problem. If that doesn't help you to figure it out,
try other variations until you do know where the problem is.

With respect to the generation of prmtops and overlapping force
fields, problems are most likely to arise if two differing atom types
are assigned the same atom type name. For example, the atom type "CG"
is sometimes used in proteins and is also used in the glycam force
field. While the two atom types might be equivalent in many cases,
this is not necessarily true. That is, the glycam force constants for
interactions involving "CG" atoms might not be appropriate for the
proteins and vice-versa. On the other hand, they might transfer just
fine -- you will have to learn enough about the force fields you are
using to figure out whether or not you might have a problem. But,
even if you do have a problem like this, it is unlikely to cause you
to have unreasonably high energies. Instead, in a situation like
this, you will most likely get energies off by only a few kcal/mol
(depending, of course, on how mis-matched the atom types are).

:-) Lachele


On Tue, Sep 9, 2008 at 9:16 AM, Waqas Nasir <nasirwaqas1983.yahoo.com> wrote:
> Hi,
>
> Hope you are fine.
>
> Well, yes I separated the 2 subunits first and then minimized all three
> files separately with same input file. I will, hopefully try and get the new
> parameter files for proteins.
>
> "you have to watch out not just for the order of prmtop generation, but
> where you load the frcmod files. I'm not sure if there is any overlap
> between ff99 and glycam, but it's something to keep in mind."
>
> I could not get the complete picture here, do you mean that the order is
> always important and moreover do you have any recommendations about loading
> frcmod files in these protein sugar cases.
>
> Thanks,
> Waqas.
>
> ----- Original Message ----
> From: Carlos Simmerling <carlos.simmerling.gmail.com>
> To: amber.scripps.edu
> Sent: Monday, September 8, 2008 7:45:51 PM
> Subject: Re: AMBER: Need help... High energies for complex...
>
> please try to be very clear about your procedures- do you mean you
> minimized all 3 systems, or minimized the complex and then directly
> separated the coordinates with no further minimization? these are
> different.
>
> you have to watch out not just for the order of prmtop generation, but
> where you load the frcmod files. I'm not sure if there is any overlap
> between ff99 and glycam, but it's something to keep in mind.
>
> finally, my advice is that you should NOT use ff99 for the protein.
> there has been a lot of discussion on this in the email reflector as
> well as in the literature.
>
>> Secondly, after minimization the DIHED values are a bit distorted, I mean
>> they do not accurately sum up to the complex but are close anyways. But
>> the
>> bond and angle values are accurate.
>> I would appreciate if somebody could comment.
>>
>> Thanks alot for the help that you people provided me with,
>> I really do appreciate that..
>> Thanks once again!
>> Regards,
>> Waqas.
>>
>>
>> ----- Original Message ----
>> From: Carlos Simmerling <carlos.simmerling.gmail.com>
>> To: amber.scripps.edu
>> Sent: Friday, September 5, 2008 12:07:01 PM
>> Subject: Re: AMBER: Need help... High energies for complex...
>>
>> no, you cannot conclude anything about binding. all you can conclude
>> is that there is a problem somewhere in your protocol. anything you
>> calculate with these will not be correct. you might want to try
>> working with a calculation that does not require editing of the
>> molecule or new parameters and see if you can get it to work on a
>> simpler test case. once you have experience with that, then you might
>> be able to get this more advanced system to work.
>>
>> On Fri, Sep 5, 2008 at 5:15 AM, Waqas Nasir <nasirwaqas1983.yahoo.com>
>> wrote:
>>> Hi,
>>>
>>> Thanks once again for such a great help.
>>>
>>> This time I have used xleap and done every thing with my hand, rather
>>> than
>>> using a script. This time the result that I have got are better but not
>>> completely correct. I am sorry for the trouble that I am causing...
>>>
>>> Here is the single point md for complex;
>>>
>>> NSTEP = 0 TIME(PS) = 0.000 TEMP(K) = 360.57 PRESS =
>>> 0.0
>>> Etot = -11596.4113 EKtot = 8487.1166 EPtot =
>>> -20083.5279
>>> BOND = 205.6103 ANGLE = 1549.3622 DIHED =
>>> 1658.6999
>>> 1-4 NB = 2829.8000 1-4 EEL = 25430.2414 VDWAALS =
>>> -3397.0994
>>> EELEC = -41648.8133 EGB = -6824.3588 RESTRAINT =
>>> 0.0000
>>> ESURF= 113.0299
>>>
>>> For sugars;
>>>
>>> NSTEP = 0 TIME(PS) = 0.000 TEMP(K) = 335.14 PRESS =
>>> 0.0
>>> Etot = 242.2702 EKtot = 79.9191 EPtot =
>>> 162.3511
>>> BOND = 15.1725 ANGLE = 47.3410 DIHED =
>>> -27.5845
>>> 1-4 NB = 26.3282 1-4 EEL = 592.4065 VDWAALS =
>>> -16.3978
>>> EELEC = -374.1894 EGB = -104.0077 RESTRAINT =
>>> 0.0000
>>> ESURF= 3.2822
>>>
>>> For proteins;
>>>
>>> NSTEP = 0 TIME(PS) = 0.000 TEMP(K) = 360.54 PRESS =
>>> 0.0
>>> Etot = -8302.0894 EKtot = 8398.2944 EPtot =
>>> -16700.3838
>>> BOND = 190.4378 ANGLE = 1485.0230 DIHED =
>>> 5266.2811
>>> 1-4 NB = 2803.4718 1-4 EEL = 24837.8348 VDWAALS =
>>> -3357.5193
>>> EELEC = -41233.4669 EGB = -6804.6020 RESTRAINT =
>>> 0.0000
>>> ESURF= 112.1559
>>>
>>> The energies are still ridiculous... but the bond energy is accurately
>>> additive where as the angles are almost. But the dihedrals are completely
>>> out. This time I have done every thing with hand so I know that the
>>> prmtop
>>> and inpcrd files are fine. What option can you think of now... to
>>> debug...
>>> should I go back to the way I created the sugars and everything... I just
>>> want to confirm a couple of things;
>>>
>>> 1. Is the protocol legitimate and proper (The way that I am calculating
>>> single point energy for each frame)?
>>> 2. Do you still think that there might be some problem in prmtop and
>>> inpcrd
>>> files (looking at the results)?
>>> 3. Do you think that these results might reflect the improper state of
>>> binding of ligand and the proteins, I mean can I conclude from these
>>> results
>>> that the sugar in question DOES NOT bind to proteins at all!!! I am
>>> running
>>> a small 200 ps md without restraints on this complex to see if the ligand
>>> remains in the binding pocket or flies off...
>>>
>>> I would really appreciate if you could comment please... I am new to
>>> amber
>>> and had just started to enjoy it until this happened...
>>>
>>> Thanks, and extremely sorry for the trouble that I am causing...
>>> Regards,
>>> Waqas.
>>>
>>> ----- Original Message ----
>>> From: Carlos Simmerling <carlos.simmerling.gmail.com>
>>> To: amber.scripps.edu
>>> Sent: Thursday, September 4, 2008 6:21:42 PM
>>> Subject: Re: AMBER: Need help... High energies for complex...
>>>
>>> if the coordinates are identical, and there is no covalent link, then
>>> the bond/angle/dihedral energies should be additive. in your case they
>>> are far from that - look at the bonds.. if there are no bonds between
>>> protein and sugar, how can the bond energies not sum up? the only
>>> conclusion I can make is that the parameters in the individual prmtop
>>> files do not match those in the prmtop for the complex, or perhaps you
>>> are not really using the same protein coordinates for the complex and
>>> isolated protein, or for complex and isolated sugar. I don't really
>>> have any way of knowing how that might have happened.
>>>
>>> On Thu, Sep 4, 2008 at 12:14 PM, Waqas Nasir <nasirwaqas1983.yahoo.com>
>>> wrote:
>>>>
>>>> The last mail that I sent you it had exactly the same coordinates in the
>>>> complex and the separate subunits. I just pasted the coordinates from
>>>> main
>>>> file in two separate files for protein and sugar subunits.
>>>>
>>>> I will re check with the prmtop and crd files but I doubt if there is
>>>> anything that I could do... its just 3 bond commands and one
>>>> saveamberparm
>>>> command. I dont have enough room to debug. And yes there is no covalent
>>>> linkage between the sugar and protein.
>>>> Do you think that its dead sure that something is different in top and
>>>> crd
>>>> files ???
>>>> Do you think, moreover, that combine command might have done something
>>>> wrong... just a thought...
>>>>
>>>> Thanks,
>>>> Waqas.
>>>>
>>>> ----- Original Message ----
>>>> From: Carlos Simmerling <carlos.simmerling.gmail.com>
>>>> To: amber.scripps.edu
>>>> Sent: Thursday, September 4, 2008 6:41:38 PM
>>>> Subject: Re: AMBER: Need help... High energies for complex...
>>>>
>>>> are the coordinates you are using for the complex exactly the same as
>>>> for the protein and sugar alone (the energies you posted earlier)?
>>>> if not, then you really can't use this to debug your energies. they
>>>> must be identical (but separate files).
>>>> if yes, then there is something wrong in the prmtop files and you will
>>>> need to go through your procedure for generating the separate files
>>>> and make sure they are the same.
>>>> all of this assumes no covalent link between sugar and protein.
>>>>
>>>> On Thu, Sep 4, 2008 at 11:22 AM, Waqas Nasir <nasirwaqas1983.yahoo.com>
>>>> wrote:
>>>>> Hi,
>>>>>
>>>>> Well, thanks a lot for the worthed help.
>>>>>
>>>>> Let me explain the whole procedure that I have taken up, I have some
>>>>> doubts
>>>>> in the way I have separated the files and the first thing that came
>>>>> into
>>>>> my
>>>>> mind after these results was exactly that the top files are not
>>>>> corresponding to what is in the results.
>>>>>
>>>>> Anyways, the sugar that I have is a tetra saccharide. I first
>>>>> constructed
>>>>> the sugar unit in xleap and added 3 bonds which were missing when I
>>>>> imported the pdb file into the xleap with glycam_06 force field (The
>>>>> reason
>>>>> being not able to find the exact residues in the glycam06 force field
>>>>> as
>>>>> are
>>>>> in the pdb file). Along side that I had a protein unit from the part of
>>>>> the
>>>>> pdb file which contained protein. I combined both units in xleap by
>>>>> combine
>>>>> command and then generated the prmtop and inpcrd files for the main
>>>>> complex
>>>>> on which I had all the minimization and md runs going on.
>>>>>
>>>>> Secondly the script for generating frames generated 250 pdb files
>>>>> corresponding to the 250 frames that I had in the trajectory. I again
>>>>> repeated the same procedure using tleap and constructed a sugar subunit
>>>>> and
>>>>> a protein subunit for each frame and generated top and crd files
>>>>> separately
>>>>> for the two subunits in each frame. These files were then used for
>>>>> subsequent single point energy calculations.
>>>>>
>>>>> Do you see mistakes in principle of the approach that has been taken up
>>>>> here???
>>>>>
>>>>> Awaiting your response,
>>>>> Thanks a lot,
>>>>> Waqas.
>>>>>
>>>>> ----- Original Message ----
>>>>> From: David A. Case <case.biomaps.rutgers.edu>
>>>>> To: amber.scripps.edu
>>>>> Sent: Thursday, September 4, 2008 5:36:37 PM
>>>>> Subject: Re: AMBER: Need help... High energies for complex...
>>>>>
>>>>> On Thu, Sep 04, 2008, Waqas Nasir wrote:
>>>>>>
>>>>>> Well, I have tried straight single point md on the complex,sugar and
>>>>>> protein
>>>>>> separately, and what I have found is as follows;
>>>>>
>>>>> This is not what you should do, at least until you understand what is
>>>>> going on. My suggestion (and Carlos'):
>>>>>
>>>>> Make a single prmtop file for the complex. Use this prmtop file for
>>>>> a single point calculation with two sets of coordinates: one with the
>>>>> ligand
>>>>> (sugar) in its binding location, and a second set of coordinates where
>>>>> you
>>>>> manually move the ligand to someplace distant from the protein, (but
>>>>> keep
>>>>> the
>>>>> protein coordinates identical to what they were in the complex.
>>>>>
>>>>> Once you see and understand how that works, you will be ready to try
>>>>> and
>>>>> interpret other experiments.
>>>>>
>>>>> ...regards...dac
>>>>>
>>>>> -----------------------------------------------------------------------
>>>>> The AMBER Mail Reflector
>>>>> To post, send mail to amber.scripps.edu
>>>>> To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
>>>>> to majordomo.scripps.edu
>>>>>
>>>>>
>>>>
>>>>
>>>>
>>>> --
>> -----------------------------------------------------------------------
>> The AMBER Mail Reflector
>> To post, send mail to amber.scripps.edu
>> To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
>> to majordomo.scripps.edu
>>
>>
>
>
>
> --
> ===================================================================
> Carlos L. Simmerling, Ph.D.
> Associate Professor Phone: (631) 632-1336
> Center for Structural Biology Fax: (631) 632-1555
> CMM Bldg, Room G80
> Stony Brook University E-mail: carlos.simmerling.gmail.com
> Stony Brook, NY 11794-5115 Web: http://comp.chem.sunysb.edu
> ===================================================================
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber.scripps.edu
> To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
> to majordomo.scripps.edu
>
>



-- 
:-) Lachele
Lachele Foley
CCRC/UGA
Athens, GA USA
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" (in the *body* of the email)
      to majordomo.scripps.edu
Received on Wed Sep 10 2008 - 06:07:44 PDT
Custom Search