- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]

From: Jason Swails <jason.swails.gmail.com>

Date: Sat, 7 Sep 2013 23:32:40 -0400

On Sat, Sep 7, 2013 at 7:21 PM, Manikanthan Bhavaraju <

manikanthanbhavaraju.gmail.com> wrote:

*> Dear All,
*

*>
*

*> I have simulated a few protein+ligand systems using an explicit solvent
*

*> model for 30 ns using ff99SB and amber12. I have calculated the binding
*

*> free energies with PBSA and GBSA using mmpbsa.py for all the systems. For
*

*> few systems, the binding energy values (both GBSA and PBSA) were in the
*

*> order of -800 kcal/mol. In order to get more insights about these values, I
*

*> am trying to calculate per-residue decomposition analysis using mmpbsa.py
*

*> script of amber12. I am using the following input file for the analysis
*

*>
*

*> GB and PB and decomp
*

*> &general
*

*> endframe=400, verbose=1,
*

*> /
*

*> &gb
*

*> igb=5, saltcon=0.100,
*

*> /
*

*> &pb
*

*> istrng=0.100,
*

*> /
*

*> &decomp
*

*> idecomp=1, print_res="1-359"
*

*> dec_verbose=1,
*

*> /
*

*> The mmpbsa.py script was successfully executed without the decomposition
*

*> analysis inputs. Upon including the corresponding terms for decomposition
*

*> analysis, I am getting the following error:
*

*>
*

It is important to note that PB and GB are not pairwise decomposable,

strictly speaking. Because energy terms in GB are pairwise additive,

however, GB can be approximately decomposed quite easily (the interaction

energies between two atoms depend on the effective radii of those atoms,

which in turn depend on other atoms). PB, on the other hand, does not have

such a clean way to approximate pairwise energies. As a result, pairwise

PB energies are very slow to calculate and may have dubious reliability.

Since GB and PB are not strictly pairwise decomposable anyway, you can

really just rely on these numbers for qualitative comparisons (or, in your

case, try and see what interactions are so heavily favoring binding). This

is my long-winded way of suggesting that you perform decomposition analysis

with just GB.

*> Loading and checking parameter files for compatibility...
*

*> Preparing trajectories for simulation...
*

*> 400 frames were processed by cpptraj for use in calculation.
*

*>
*

*> Running calculations on normal system...
*

*>
*

*> Beginning GB calculations with /usr/local/amber12/bin/sander
*

*> calculating complex contribution...
*

*> bad atom type: C0
*

*>
*

This seems like a strange error to me. It comes from line ca. 2127 of

$AMBERHOME/src/sander/mdread.F90 in Amber 12, and does not occur when not

using decomposition because a non-decomposition simulations use a different

model for surface area calculations (LCPO vs. ICOSA for decomp). However,

this parameter assignment should be done based on the atomic number (check

that "%FLAG ATOMIC_NUMBER" exists in your complex.prmtop file). I'm

assuming that C0 is a carbon atom, yes? If so, this error should not occur

unless there is something weird with your system setup (did you use

AmberTools 12 or 13 to build your prmtop? Is this atom not really a

carbon?). Also, the C0 atom type does not sound familiar in the context of

ff99SB...

HTH,

Jason

Date: Sat, 7 Sep 2013 23:32:40 -0400

On Sat, Sep 7, 2013 at 7:21 PM, Manikanthan Bhavaraju <

manikanthanbhavaraju.gmail.com> wrote:

It is important to note that PB and GB are not pairwise decomposable,

strictly speaking. Because energy terms in GB are pairwise additive,

however, GB can be approximately decomposed quite easily (the interaction

energies between two atoms depend on the effective radii of those atoms,

which in turn depend on other atoms). PB, on the other hand, does not have

such a clean way to approximate pairwise energies. As a result, pairwise

PB energies are very slow to calculate and may have dubious reliability.

Since GB and PB are not strictly pairwise decomposable anyway, you can

really just rely on these numbers for qualitative comparisons (or, in your

case, try and see what interactions are so heavily favoring binding). This

is my long-winded way of suggesting that you perform decomposition analysis

with just GB.

This seems like a strange error to me. It comes from line ca. 2127 of

$AMBERHOME/src/sander/mdread.F90 in Amber 12, and does not occur when not

using decomposition because a non-decomposition simulations use a different

model for surface area calculations (LCPO vs. ICOSA for decomp). However,

this parameter assignment should be done based on the atomic number (check

that "%FLAG ATOMIC_NUMBER" exists in your complex.prmtop file). I'm

assuming that C0 is a carbon atom, yes? If so, this error should not occur

unless there is something weird with your system setup (did you use

AmberTools 12 or 13 to build your prmtop? Is this atom not really a

carbon?). Also, the C0 atom type does not sound familiar in the context of

ff99SB...

HTH,

Jason

-- Jason M. Swails BioMaPS, Rutgers University Postdoctoral Researcher _______________________________________________ AMBER mailing list AMBER.ambermd.org http://lists.ambermd.org/mailman/listinfo/amberReceived on Sat Sep 07 2013 - 21:00:02 PDT

Custom Search