Biological structure optimization issue

[UlysseFerier]

XTB.txt 1KF1_optimization_Marc.txt 1KF1_Na+.txt

We have issues optimizing a biological structure with Na+ counterions (PDB entry : 1KF1). XTB version : XTB/6.7.0

The structure was successfully optimized in its -18 charged form, in GBSA (H2O) / GFN2 / opt loose / chrg -18 (see 1KF1_optimization_Marc.txt for the optimized structure)

Na+ counterions were manually placed on the optimized structure, in front of each phosphate group and we ran an optimization for this new structure (net total charge is now +3, 21 Na+ added).

The command line used for this optimization was the following : xtb 1KF1_Na+.xyz --gbsa h2o --verbose --opt loose --chrg 3 The input file is 1KF1_Na+.txt and the XTB.txt file contains the results.

After the 114th cycle of optimization, no energy gain could be found for each step. The software crashes and repeats the same operation until the 1886th cycle :

........................................................................
.............................. CYCLE  114 ..............................
........................................................................
 * RMSD in coord.:     0.0151464 a0     energy gain   0.1815320E+04 Eh
 * calculating model hessian...
Using Lindh-Hessian (1995)
 Shifting diagonal of input Hessian by   1.023591329227202E-002
 Lowest  eigenvalues of input Hessian
    0.010000    0.010064    0.010117    0.010169    0.010171    0.010188
    0.010192    0.010196    0.010201    0.010201    0.010202    0.010210
    0.010222    0.010223    0.010224    0.010227    0.010232    0.010233
 Highest eigenvalues
    3.187131    3.221799    3.278592    3.303234    3.338850    3.488041

........................................................................
.............................. CYCLE  115 ..............................
........................................................................
 * RMSD in coord.:     0.0000007 a0     energy gain   0.0000000E+00 Eh
 * calculating model hessian...
Using Lindh-Hessian (1995)
 Shifting diagonal of input Hessian by   1.023591329227202E-002
 Lowest  eigenvalues of input Hessian
    0.010000    0.010064    0.010117    0.010169    0.010171    0.010188
    0.010192    0.010196    0.010201    0.010201    0.010202    0.010210
    0.010222    0.010223    0.010224    0.010227    0.010232    0.010233
 Highest eigenvalues
    3.187131    3.221799    3.278592    3.303234    3.338850    3.488041

........................................................................
.............................. CYCLE  116 ..............................
........................................................................
 * RMSD in coord.:     0.0000007 a0     energy gain   0.0000000E+00 Eh
 * calculating model hessian...
Using Lindh-Hessian (1995)
 Shifting diagonal of input Hessian by   1.023591329227202E-002
 Lowest  eigenvalues of input Hessian
    0.010000    0.010064    0.010117    0.010169    0.010171    0.010188
    0.010192    0.010196    0.010201    0.010201    0.010202    0.010210
    0.010222    0.010223    0.010224    0.010227    0.010232    0.010233
 Highest eigenvalues
    3.187131    3.221799    3.278592    3.303234    3.338850    3.488041

.
.
........................................................................
.............................. CYCLE 1885 ..............................
........................................................................
 * RMSD in coord.:     0.0000007 a0     energy gain   0.0000000E+00 Eh
 * calculating model hessian...
Using Lindh-Hessian (1995)
 Shifting diagonal of input Hessian by   1.023591329227202E-002
 Lowest  eigenvalues of input Hessian
    0.010000    0.010064    0.010117    0.010169    0.010171    0.010188
    0.010192    0.010196    0.010201    0.010201    0.010202    0.010210
    0.010222    0.010223    0.010224    0.010227    0.010232    0.010233
 Highest eigenvalues
    3.187131    3.221799    3.278592    3.303234    3.338850    3.488041
 ........................................................................
.............................. CYCLE 1886 ..............................
........................................................................
 * RMSD in coord.:     0.0000007 a0     energy gain   0.0000000E+00 Eh

   *** FAILED TO CONVERGE GEOMETRY OPTIMIZATION IN 1886 CYCLES ***

Finally, many errors can be found at the end of the XTB.log file such as :

########################################################################
[ERROR] Program stopped due to fatal error
-7098- Geometry optimization failed
-7097- xtb_geoopt: Trying to recover from failed geometry optimization
-7096- xtb_geoopt: Geometry optimization did not converge
-7095- relaxation_engine_lbfgs_relax: SCF not converged, aborting...
-7094- xtb_calculator_singlepoint: Electronic structure method terminated
-7093- scf: Setup of Coulomb evaluator failed
-7092- type_latticepoint_update: Could not generate lattice points
.
.
-4- type_latticepoint_update: Could not generate lattice points
-3- relaxation_engine_lbfgs_relax: SCF not converged, aborting...
-2- xtb_calculator_singlepoint: Electronic structure method terminated
-1- scf: Self consistent charge iterator did not converge
########################################################################

[cplett]

Hi, It seems that the geometry optimization led to a structure where the SCC didn't converge anymore. You might check the optimization path in the xtbopt.xyz file. As this might be a GFN2-xTB problem, I would recommend trying GFN1-xTB.

[marcelmbn]

@cplett This looks more like the geometry optimizer couldn't converge due to technical reasons, doesn't it?

[mdewergi]

This is what I was thinking and want you to check for... GFN1 is failing as well.