Re: [AMBER] Problem with mdnab: ERROR in RATTLE

From: Andrey <>
Date: Tue, 17 Apr 2012 18:20:42 +0400

On 04/14/2012 10:04 PM, case wrote:
> On Fri, Apr 13, 2012, Andrey wrote:
>> However, when I try to run subsequent equilibrium with mdnab, for 2 of 8
>> complexes I encounter the following error:
>> mdnab "../min/pdb/${PREFIX}.pdb"
>> "../../build/amber/${PREFIX}.leap.prm" "traj/${PREFIX}" 2 500
>> Reading parm file (../../build/amber/cdk6_p6_3_cplx.leap.prm)
>> iter Total bad vdW elect nonpolar
>> genBorn frms
>> ff: 0 -7285.90 4577.15 -1029.51 -6770.49 0.00 -4063.05
>> 3.53e-01
>> For other complexes besides these two (even the ones with the same
>> peptide, but another binding point) mdnab finishes successfully.
> This is indeed odd. I'm assuming that the nab tests (especially
> Run.rattle)
> passed. You might try a test run in dt=0.001 to see if that helps. Also
> set ntpr_md=1 so that you can see the most information (just run a few
> steps,
> since it basically goes haywire at the first step.)
David, thanks for your reply.

All test performed by `make test` pass.

Running my complex at dt=0.001 (and even 0.0001 and 0.000001) changes
nothing, it immediately NaNs.
Running with ntpr_md=1 shows that it fails at the very first step:
         mm_options: cut=12.000000
         mm_options: nsnb=25
         mm_options: diel=C
         mm_options: gb=2
         mm_options: rgbmax=12.000000
         mm_options: rattle=1
         mm_options: dt=0.000100
         mm_options: ntpr=101
         mm_options: ntpr_md=1
         mm_options: ntwx=0
         mm_options: zerov=0
         mm_options: tempi=50.000000
         mm_options: temp0=100.000000
         mm_options: gamma_ln=20.000000
         mm_options: wcons=0.000000
       iter Total bad vdW elect nonpolar
genBorn frms
ff: 0 -7285.94 4576.65 -1028.96 -6771.99 0.00 -4061.63
md: 1 0.000 -nan -nan -nan -nan
md: 2 0.000 -nan -nan -nan -nan

> Otherwise, I think you will have to post examples of a job that works
> and one
> that fails. The minnab and mdnab codes are fairly new, and only a few
> people
> are using them, so there could well be hidden problems lurking.
> Also let us know what version of AmberTools you are using, and which
> compiler
> and OS.
I've used two machines, both give the same result.
AmberTools12, gcc 4.6.1 on Ubuntu 11.10 x86_64
AmberTools12, gcc 4.4.3 on Ubuntu 10.04 x86_64
./configure -rism gnu

I've tried changing `maxiter` parameter in `rattle` function to 100000
(instead of 1000): it still NaNs at the first step of MD, but does not
produce any error.

An archive with .prm/.crd/.pdb files produced by pytleap and minimized
.pdb files (in min/ directory) is available at
The parameters are `mdnab <minimized.pdb> <param.prm> <outfile> 2 500`
cdk6_p6_3 is faulty one, cdk6_p6_1 runs perfectly fine.

Another question: If I disable RATTLE at all and then run simulation
with dt=0.001, will it be enough to produce sensible results, or will
disabling RATTLE require more changes to MD parameters?

Thanks in advance,

Andrey Alekseenko
Moscow Institute of Physics and Technology
> ....thanks...dac
> _______________________________________________
> AMBER mailing list

AMBER mailing list
Received on Tue Apr 17 2012 - 07:30:04 PDT
Custom Search