LINCS warning during energy minimization

User discussions
1

GROMACS version: 2022.3
GROMACS modification: No

Hi everyone,
I am running a simulation of a protein preparing the structure in CHARMM-GUI. I used the mdp file obtained by CHARMM-GUI but during the energy minimization it shows the following warning:

Steepest Descents:
Tolerance (Fmax) = 1.00000e+03
Number of steps = 5000
Step= 0, Dmax= 1.0e-02 nm, Epot= 4.64660e+08 Fmax= 3.88728e+10, atom= 16241
Step= 1, Dmax= 1.0e-02 nm, Epot= 1.11108e+08 Fmax= 1.33617e+10, atom= 16241
Step= 2, Dmax= 1.2e-02 nm, Epot= 1.60567e+07 Fmax= 2.20310e+09, atom= 16241
Step= 3, Dmax= 1.4e-02 nm, Epot= 1.29812e+06 Fmax= 2.52864e+08, atom= 16241
Step= 4, Dmax= 1.7e-02 nm, Epot= -1.39038e+06 Fmax= 2.01259e+07, atom= 4091
Step= 5, Dmax= 2.1e-02 nm, Epot= -1.87866e+06 Fmax= 2.62460e+06, atom= 16240
Step= 6, Dmax= 2.5e-02 nm, Epot= -1.95552e+06 Fmax= 5.41096e+05, atom= 16241
Step= 7, Dmax= 3.0e-02 nm, Epot= -1.97906e+06 Fmax= 1.31029e+05, atom= 16241
Step= 8, Dmax= 3.6e-02 nm, Epot= -1.98812e+06 Fmax= 5.19950e+04, atom= 16241
Step= 9, Dmax= 4.3e-02 nm, Epot= -1.99386e+06 Fmax= 1.89779e+04, atom= 16545
Step= 10, Dmax= 5.2e-02 nm, Epot= -2.00012e+06 Fmax= 1.27571e+05, atom= 16545
Step= 11, Dmax= 6.2e-02 nm, Epot= -2.00363e+06 Fmax= 1.13476e+04, atom= 16240
Step= 12, Dmax= 7.4e-02 nm, Epot= -2.00986e+06 Fmax= 9.47107e+04, atom= 16240
Step= 13, Dmax= 8.9e-02 nm, Epot= -2.01693e+06 Fmax= 5.51797e+04, atom= 16240
Step= 14, Dmax= 1.1e-01 nm, Epot= -2.01999e+06 Fmax= 4.25321e+04, atom= 16240

Step 15, time 0.015 (ps) LINCS WARNING
relative constraint deviation after LINCS:
rms 0.000016, max 0.000952 (between atoms 16243 and 16246)
bonds that rotated more than 30 degrees:
atom 1 atom 2 angle previous, current, constraint length
16227 16228 35.6 0.1090 0.1090 0.1090
Step= 16, Dmax= 6.4e-02 nm, Epot= -2.02320e+06 Fmax= 3.54836e+04, atom= 16240
Step= 17, Dmax= 7.7e-02 nm, Epot= -2.02465e+06 Fmax= 6.53873e+04, atom= 16240
Step= 18, Dmax= 9.2e-02 nm, Epot= -2.02726e+06 Fmax= 4.97583e+04, atom= 16240
Step= 20, Dmax= 5.5e-02 nm, Epot= -2.03076e+06 Fmax= 2.10320e+04, atom= 16240
Step= 21, Dmax= 6.7e-02 nm, Epot= -2.03132e+06 Fmax= 5.81503e+04, atom= 16240
Step= 22, Dmax= 8.0e-02 nm, Epot= -2.03454e+06 Fmax= 4.07968e+04, atom= 16240
Step= 24, Dmax= 4.8e-02 nm, Epot= -2.03758e+06 Fmax= 1.92212e+04, atom= 16240
Step= 26, Dmax= 2.9e-02 nm, Epot= -2.03906e+06 Fmax= 1.60387e+04, atom= 16240

Here is the mdp file I used:

define = -DPOSRES -DPOSRES_FC_BB=400.0 -DPOSRES_FC_SC=40.0
integrator = steep
emtol = 1000.0
nsteps = 5000
emstep = 0.01
nstlist = 10
cutoff-scheme = Verlet
rlist = 0.9
vdwtype = Cut-off
vdw-modifier = None
DispCorr = EnerPres
rvdw = 0.9
coulombtype = PME
rcoulomb = 0.9
;
constraints = h-bonds
constraint_algorithm = LINCS

Is this warning related to something wrong in my structure that I need to solve?
Is it correct if I simply change the emstep to 0,005 (because I am quite new to gromacs and I have never found a value smaller than 0,01.)
Otherwise I saw that I can avoid that lincs warning by changing the lincs-warnangle to 40 but I think this is a more important change than only changing the emstep.
So, I was wondering which is the more correct way to solve the problem?

Thank you very much

2

You can looks around the atoms implicated (circa 16240-16246 from the log) in your initial structure, and check for obvious problems in a visualization software.

LINCS warnings are indeed indicative of a structure problem when they occur during a simulation. But seeing one or two during the first steps of energy minimization (as here) is not too worrying.

Changing the emsteps would be okay, but possibly unecessary if that is the only LINCS warning. Changing the lincs-warnangle is inadvisable (masking rather than solving the problem).

3

The CHARMM-GUI protocol also involves gradually releasing restraints, so in this case where several types of restraints are used simultaneously, they may be impeding progress toward the minimum. Subsequent minimizations with weaker (and ultimately no) restraints should complete smoothly. If not, there is something more fundamentally wrong with the system.