[gmx-users] Umbrella sampling simulations using make_edi with -restrain and -harmonic

[Chen, Po-chia]

 Hi all,
 I can't seem to find the correct combination of EDI parameters impose a harmonically-constrained simulation along an eigenvector in GROMACS 4.6.2, and I'd like to confirm that the inputs I have are actually correct. Here is the make_edi command I used to generate the .edi file fed to mdrun:

echo "C-alpha System" | make_edi -restrain -harmonic \
 -f ../ca-evec.trr \
 -eig ../ca-eval.xvg \
 -s ../analysis.tpr \
 -ori ./init.gro \
 -outfrq 2500 -deltaF0 150 -Eflnull 100 -flood 1 \
 -o constrain.edi

 Where the eigenvectors and eigenvalues are previously derived by g_covar for a set of unrestrained trajectories. The origin file init.gro is the same as the starting coordinates of this constrained run, extracted from a previous EDI run to start at an appropriate location on the eigenvector space.

 The 1st eigenvalue is ~ 66, and the eigenvector looked fine in VMD when plotted by g_anaeig -extr.

 So mdrun -ei constrain.edi runs normally, but the flooding potential drops to zero so the protein diffuses freely along the 1st eigenvector. e.g. the first/last line in edsam.xvg looks like:
# time RMSD EV1projFLOOD EV1-Ef1  EV1-Vf1 EV1-deltaF
  0.0   0.43   9.99   -9.4e+1 9.5e-8  3.8e-9  -5.3e-4
 10.0  0.40   9.49  0.0   -0.0  1.6e-44    -0.0
 ...
  10000.00 0.94  -2.2 0.0 -0.0 1.6e-44    -0.0

 When c(0) is supposed to be according 9.99 as I set init.gro to be the -ori. Do I need to change -alpha as well? What parameters am I missing/added by mistake? The manual gives no indication as to which ones to do.

 = = =
 P.S. the relevant constrain.edi lines contain:
...
#DELTA_F0
150.00
#INIT_DELTA_F
0.0
#TAU
0.10
#EFL_NULL
-100.00
#ALPHA2
-1.0
#KT
2.50
#HARMONIC
1
#CONST_FORCE_FLOODING
0
...