- Posts: 8
Segmentation fault
- cchan2242
-
Topic Author
- Offline
- Fresh Boarder
Less
More
9 years 6 months ago #4115
by cchan2242
Segmentation fault was created by cchan2242
Hi all,
I have been struggling simulation of a protein above a lipid bilayer. I found it very difficult to successfully finish even the equilibration. Following is my mdp options:
define = -DPOSRES
dt = 0.02
nsteps = 25000
nstxout = 0
nstvout = 0
nstlog = 100
nstxtcout = 100
xtc-precision = 10
rlist = 1.4
coulombtype = shift
rcoulomb = 1.2
epsilon_r = 15
vdw-type = shift
rvdw-switch = 0.9
rvdw = 1.2
tcoupl = v-rescale
tc-grps = Protein Lipids Solvent
tau-t = 1.0 1.0 1.0
ref-t = 300 300 300
Pcoupl = parrinello-rahman
Pcoupltype = semiisotropic
tau-p = 12.0 12.0
compressibility = 3e-4 3e-4
ref-p = 1.0 1.0
refcoord_scaling = all
The system consists of a protein at about 2 nm above a lipid bilayer of 3 components (DOPC, DOPE and POPS).
To have a more stable starting structure, I have equilibrated the protein for several tens of ns and then use insane.py to add the membrane and solvent. Also I have run quite a long minimization and ensure the max force inside the system is less than 20 kcal/mol.
However I am still not able to go further with the equilibration and Gromacs returns segmentation fault.
I will greatly appreciate that if anyone can share his experience on this.
I have been struggling simulation of a protein above a lipid bilayer. I found it very difficult to successfully finish even the equilibration. Following is my mdp options:
define = -DPOSRES
dt = 0.02
nsteps = 25000
nstxout = 0
nstvout = 0
nstlog = 100
nstxtcout = 100
xtc-precision = 10
rlist = 1.4
coulombtype = shift
rcoulomb = 1.2
epsilon_r = 15
vdw-type = shift
rvdw-switch = 0.9
rvdw = 1.2
tcoupl = v-rescale
tc-grps = Protein Lipids Solvent
tau-t = 1.0 1.0 1.0
ref-t = 300 300 300
Pcoupl = parrinello-rahman
Pcoupltype = semiisotropic
tau-p = 12.0 12.0
compressibility = 3e-4 3e-4
ref-p = 1.0 1.0
refcoord_scaling = all
The system consists of a protein at about 2 nm above a lipid bilayer of 3 components (DOPC, DOPE and POPS).
To have a more stable starting structure, I have equilibrated the protein for several tens of ns and then use insane.py to add the membrane and solvent. Also I have run quite a long minimization and ensure the max force inside the system is less than 20 kcal/mol.
However I am still not able to go further with the equilibration and Gromacs returns segmentation fault.
I will greatly appreciate that if anyone can share his experience on this.
Please Log in or Create an account to join the conversation.
- xavier
-
- Offline
- Admin
Less
More
- Posts: 416
9 years 6 months ago #4116
by xavier
Replied by xavier on topic Segmentation fault
For equilibration purpose you want to switch the pressure compiling to a Berendsen barostat with a tau-p of ~5 ps.
This should help.
This should help.
Please Log in or Create an account to join the conversation.
- cchan2242
-
Topic Author
- Offline
- Fresh Boarder
Less
More
- Posts: 8
9 years 6 months ago #4120
by cchan2242
Replied by cchan2242 on topic Segmentation fault
The suggestion works fine. Thanks so much.
Please Log in or Create an account to join the conversation.
- doa.hawamdeh
-
- Offline
- Fresh Boarder
Less
More
- Posts: 18
9 years 4 months ago #4233
by doa.hawamdeh
Replied by doa.hawamdeh on topic Segmentation fault
Hi,
I have the same problem, but this suggestion did not work with me !! Is there any other suggestions??
My system consists of two large molecules one has a positive charge and the other has a negative charge, and I used shifted potential for the electrostatic interactions!!
Thank you.
I have the same problem, but this suggestion did not work with me !! Is there any other suggestions??
My system consists of two large molecules one has a positive charge and the other has a negative charge, and I used shifted potential for the electrostatic interactions!!
Thank you.
Please Log in or Create an account to join the conversation.
- xavier
-
- Offline
- Admin
Less
More
- Posts: 416
9 years 3 months ago #4241
by xavier
Replied by xavier on topic Segmentation fault
Why do you use shift potentials?
You system might not be well equilibrated …
You system might not be well equilibrated …
Please Log in or Create an account to join the conversation.
- doa.hawamdeh
-
- Offline
- Fresh Boarder
Less
More
- Posts: 18
9 years 3 months ago #4244
by doa.hawamdeh
Replied by doa.hawamdeh on topic Segmentation fault
What I must use instead of shift potentials?
Please Log in or Create an account to join the conversation.
- mnmelo
-
- Offline
- Admin
Less
More
- Posts: 89
9 years 3 months ago #4250
by mnmelo
Replied by mnmelo on topic Segmentation fault
Hi,
Shift potentials are ok, if you use them the same way cchan does (which is the Martini way).
There are a number of other potential reasons why your system fails to equilibrate:
Did you also use insane.py to build your system, or did you do it some other way?
Did you properly energy-minimize the solvated protein+bilayer structure properly before trying to run it? What kind of energies and forces did you get on your last minimization step?
Sharing your .mdp might help us help you, too. (if you copy off cchan's .mdp, beware that a xtc-precision of 10 is probably too low even for Martini; go for 100).
Cheers,
Manel
Shift potentials are ok, if you use them the same way cchan does (which is the Martini way).
There are a number of other potential reasons why your system fails to equilibrate:
Did you also use insane.py to build your system, or did you do it some other way?
Did you properly energy-minimize the solvated protein+bilayer structure properly before trying to run it? What kind of energies and forces did you get on your last minimization step?
Sharing your .mdp might help us help you, too. (if you copy off cchan's .mdp, beware that a xtc-precision of 10 is probably too low even for Martini; go for 100).
Cheers,
Manel
Please Log in or Create an account to join the conversation.
- doa.hawamdeh
-
- Offline
- Fresh Boarder
Less
More
- Posts: 18
9 years 3 months ago #4253
by doa.hawamdeh
Replied by doa.hawamdeh on topic Segmentation fault
Hi
My system consists of generation 4 of PAMAM dendrimer (250 beads) and double stranded dna(177 beads) {MARTINI forcefield}. Regarding the dendrimer I used the same model that Lee and Larson used(H. Lee and R. G. Larson, The Journal of Physical Chemistry B 110 (37), 18204-18211 (2006)), and I used $ python martinize-dna.py –dnatype ds –f 2M8Z.pdb -o cg-dna.top –x cg-dna.pdb to create the coarse-grained topology of the DNA.
then I solvated the system and equilibrated it.Below is the parameters that was used to minimize the system.
integrator = steep
dt = 0.02
nsteps = 30000
nstxout = 0
nstvout = 0
nstfout = 0
nstlog = 1000
nstenergy = 100
nstxtcout = 1000
xtc_precision = 100
nstlist = 10 ; 10 fs
ns_type = grid ; search neighboring grid cells
pbc = xyz
rlist =1.4 ; short-range neighborlist cutoff (in nm)
coulombtype = shift ; Particle Mesh Ewald for long-range electrostatics
;fourierspacing = 0.12 ; grid spacing for FFT
;pme_order = 4 ; cubic interpolation shoould give electrostatic ;energies accurate to about 5 e-3.
;optimize_fft = yes
rcoulomb_switch = 0.0
rcoulomb = 1.2
epsilon_r = 15
vdw_type = Shift
rvdw_switch = 0.9
rvdw = 1.2
;ewald_rtol =1e-5
;cutoff-scheme =group
tcoupl = v-rescale
tc-grps = System
tau_t = 1.0
ref_t = 300
Pcoupl = parrinello-rahman
Pcoupltype = isotropic
tau_p = 5.0
compressibility = 3e-4
ref_p = 1.0
gen_vel = no
gen_temp = 310
gen_seed = 47523
constraints = none
constraint_algorithm = Lincs
unconstrained_start = no
lincs_order = 4
lincs_warnangle = 30
Thank you so much
Doa
My system consists of generation 4 of PAMAM dendrimer (250 beads) and double stranded dna(177 beads) {MARTINI forcefield}. Regarding the dendrimer I used the same model that Lee and Larson used(H. Lee and R. G. Larson, The Journal of Physical Chemistry B 110 (37), 18204-18211 (2006)), and I used $ python martinize-dna.py –dnatype ds –f 2M8Z.pdb -o cg-dna.top –x cg-dna.pdb to create the coarse-grained topology of the DNA.
then I solvated the system and equilibrated it.Below is the parameters that was used to minimize the system.
integrator = steep
dt = 0.02
nsteps = 30000
nstxout = 0
nstvout = 0
nstfout = 0
nstlog = 1000
nstenergy = 100
nstxtcout = 1000
xtc_precision = 100
nstlist = 10 ; 10 fs
ns_type = grid ; search neighboring grid cells
pbc = xyz
rlist =1.4 ; short-range neighborlist cutoff (in nm)
coulombtype = shift ; Particle Mesh Ewald for long-range electrostatics
;fourierspacing = 0.12 ; grid spacing for FFT
;pme_order = 4 ; cubic interpolation shoould give electrostatic ;energies accurate to about 5 e-3.
;optimize_fft = yes
rcoulomb_switch = 0.0
rcoulomb = 1.2
epsilon_r = 15
vdw_type = Shift
rvdw_switch = 0.9
rvdw = 1.2
;ewald_rtol =1e-5
;cutoff-scheme =group
tcoupl = v-rescale
tc-grps = System
tau_t = 1.0
ref_t = 300
Pcoupl = parrinello-rahman
Pcoupltype = isotropic
tau_p = 5.0
compressibility = 3e-4
ref_p = 1.0
gen_vel = no
gen_temp = 310
gen_seed = 47523
constraints = none
constraint_algorithm = Lincs
unconstrained_start = no
lincs_order = 4
lincs_warnangle = 30
Thank you so much
Doa
Please Log in or Create an account to join the conversation.
- doa.hawamdeh
-
- Offline
- Fresh Boarder
Less
More
- Posts: 18
9 years 3 months ago #4254
by doa.hawamdeh
Replied by doa.hawamdeh on topic Segmentation fault
Sorry I forgot to mention the value of the potential energy at the last minimization step:
Potential Energy = -3.3400072e+05
Maximum force = 6.2834332e+02 on atom 326
Norm of force = 1.0440180e+01
after minimization I ran md (time step=20 fs) but I obtained segmentation fault.
Waiting your answer.
Thank you.
Doa
Potential Energy = -3.3400072e+05
Maximum force = 6.2834332e+02 on atom 326
Norm of force = 1.0440180e+01
after minimization I ran md (time step=20 fs) but I obtained segmentation fault.
Waiting your answer.
Thank you.
Doa
Please Log in or Create an account to join the conversation.
- mnmelo
-
- Offline
- Admin
Less
More
- Posts: 89
9 years 3 months ago #4264
by mnmelo
Replied by mnmelo on topic Segmentation fault
Hi,
Your minimization parameters seem ok (though most of your.mdp lines --the ones pertaining to temperature and pressure coupling-- are ignored for the minimization). I guess your problem lies with the equilibration.
If you're using for equilibration the same sort of pressure coupling settings as you showed, then you didn't follow the first advice Xavier gave in this post: use the Berendsen barostat for equilibration.
And even if afterwards you move to parinello-rahman your tau-p should be larger (closer to 12ps). Otherwise you might end up with too-high fluctuations in your box size.
Cheers,
Manel
Your minimization parameters seem ok (though most of your.mdp lines --the ones pertaining to temperature and pressure coupling-- are ignored for the minimization). I guess your problem lies with the equilibration.
If you're using for equilibration the same sort of pressure coupling settings as you showed, then you didn't follow the first advice Xavier gave in this post: use the Berendsen barostat for equilibration.
And even if afterwards you move to parinello-rahman your tau-p should be larger (closer to 12ps). Otherwise you might end up with too-high fluctuations in your box size.
Cheers,
Manel
Please Log in or Create an account to join the conversation.
- doa.hawamdeh
-
- Offline
- Fresh Boarder
Less
More
- Posts: 18
9 years 3 months ago #4288
by doa.hawamdeh
Replied by doa.hawamdeh on topic Segmentation fault
Thank you for reply.
The equilibration did not work with me :( I tried both suggestions but I still get "segmentation fault (core dumped)"
This is the equilibration parameter file:
define = -DPOSRES
dt = 0.02
nsteps = 25000
nstxout = 0
nstvout = 0
nstlog = 100
nstxtcout = 100
xtc_precision = 10
nstlist = 10
ns_type = grid
pbc = xyz
rlist =1.4
coulombtype = shift
rcoulomb = 1.2
epsilon_r = 15
vdw_type = Shift
rvdw_switch = 0.9
rvdw = 1.2
cutoff-scheme =group
tcoupl = v-rescale
tc-grps = system
tau_t = 1.0
ref_t = 310
Pcoupl = parrinello-rahman
Pcoupltype = isotropic
tau_p = 12.0
compressibility = 3e-4
ref_p = 1.0
refcoord_scaling = all
Also I tried berendsen barostat for pressure coupling with coupling constant 5.0 and the result is segmentation fault!!!!!!!!!
Doa
The equilibration did not work with me :( I tried both suggestions but I still get "segmentation fault (core dumped)"
This is the equilibration parameter file:
define = -DPOSRES
dt = 0.02
nsteps = 25000
nstxout = 0
nstvout = 0
nstlog = 100
nstxtcout = 100
xtc_precision = 10
nstlist = 10
ns_type = grid
pbc = xyz
rlist =1.4
coulombtype = shift
rcoulomb = 1.2
epsilon_r = 15
vdw_type = Shift
rvdw_switch = 0.9
rvdw = 1.2
cutoff-scheme =group
tcoupl = v-rescale
tc-grps = system
tau_t = 1.0
ref_t = 310
Pcoupl = parrinello-rahman
Pcoupltype = isotropic
tau_p = 12.0
compressibility = 3e-4
ref_p = 1.0
refcoord_scaling = all
Also I tried berendsen barostat for pressure coupling with coupling constant 5.0 and the result is segmentation fault!!!!!!!!!
Doa
Please Log in or Create an account to join the conversation.
- mnmelo
-
- Offline
- Admin
Less
More
- Posts: 89
9 years 3 months ago #4449
by mnmelo
Replied by mnmelo on topic Segmentation fault
Hello again, Happy 2015!
Purely from the .mdp perspective the simulations should be running fine (I'd change xtc_precision to at least 100, but this won't affect the run at all, just the output). I notice you're using position restraints. Maybe the restraining forces are too high? Also, if restraining a protein, apply the restraining potentials only on the backbone beads.
If you want more specific help you'll have to be more specific about the system: what it is and where you got the topology from. In the meanwhile here is some more generic advice: besides the run parameters (.mdp) the two other sources of problems can be a)the starting configuration, b)the topology.
a) Usually a configuration that energy-minimizes to low energies/forces is stable enough to run. However, in some cases this might not be enough to correct overlaps (for example, if you simply overlap a protein into a membrane without removing any lipids). In this case you must find the sources of clashes, fix them (may involve resolvating the system), and minimize again.
b) If you're creating a new topology it is possible that you included potentials that are too stiff, or that become numerically unstable under certain conditions. If this is the case check our tutorials on parameterizing new molecules for hints on where to go from here.
Besides these tips, check the logs and the output of mdrun. Often there will be several indications of problems before a segfault. Check the beginning of the log for the "Initial maximum inter charge-group distances". If any of these is greater than a couple nm then you have something broken in your structure.
Finally, a good way to spot where in your system things begin to explode is to save every simulation frame (nstxout = 1). You then get to see a super slow motion movie of your system right before it becomes unstable.
Cheers and let me know if any of this helps,
Manel
Purely from the .mdp perspective the simulations should be running fine (I'd change xtc_precision to at least 100, but this won't affect the run at all, just the output). I notice you're using position restraints. Maybe the restraining forces are too high? Also, if restraining a protein, apply the restraining potentials only on the backbone beads.
If you want more specific help you'll have to be more specific about the system: what it is and where you got the topology from. In the meanwhile here is some more generic advice: besides the run parameters (.mdp) the two other sources of problems can be a)the starting configuration, b)the topology.
a) Usually a configuration that energy-minimizes to low energies/forces is stable enough to run. However, in some cases this might not be enough to correct overlaps (for example, if you simply overlap a protein into a membrane without removing any lipids). In this case you must find the sources of clashes, fix them (may involve resolvating the system), and minimize again.
b) If you're creating a new topology it is possible that you included potentials that are too stiff, or that become numerically unstable under certain conditions. If this is the case check our tutorials on parameterizing new molecules for hints on where to go from here.
Besides these tips, check the logs and the output of mdrun. Often there will be several indications of problems before a segfault. Check the beginning of the log for the "Initial maximum inter charge-group distances". If any of these is greater than a couple nm then you have something broken in your structure.
Finally, a good way to spot where in your system things begin to explode is to save every simulation frame (nstxout = 1). You then get to see a super slow motion movie of your system right before it becomes unstable.
Cheers and let me know if any of this helps,
Manel
Please Log in or Create an account to join the conversation.
- mnmelo
-
- Offline
- Admin
Less
More
- Posts: 89
9 years 3 months ago #4450
by mnmelo
Replied by mnmelo on topic Segmentation fault
I apologize, I missed the post where you already fully describe your system.
In your case it may be that the topologies you are using don't play very well with each other, although I wouldn't expect it to explode in such a short equilibration time.
Are you using your dendrimer as the solvent? Or is there water in the system? In the former case it might be tricky to get a proper solvation without too many overlaps.
Try the tips I wrote in the last post (especially the output at every step) and let me know.
Cheers,
Manel
In your case it may be that the topologies you are using don't play very well with each other, although I wouldn't expect it to explode in such a short equilibration time.
Are you using your dendrimer as the solvent? Or is there water in the system? In the former case it might be tricky to get a proper solvation without too many overlaps.
Try the tips I wrote in the last post (especially the output at every step) and let me know.
Cheers,
Manel
Please Log in or Create an account to join the conversation.
- doa.hawamdeh
-
- Offline
- Fresh Boarder
Less
More
- Posts: 18
9 years 2 months ago #4452
by doa.hawamdeh
Replied by doa.hawamdeh on topic Segmentation fault
Thank you so much for your ttips. I apologies for late reply.
I have followed your tips but I did not get positive results, I still have the same problem.
The dendrimer I use is not a solvent, I solvate both of dendrimer and DNA in water. I simulated the dendrimer alone in water and there was no problem everything went will, but the segmentation fault problem appears when I want to equilibrate the DNA. I downloaded the pdb file of the all atom DNA structure from the pdb.org and used the python script (martinize-dna.py) to convert the all atom to coarse-grained. I have a doubt that the problem is in the structure!! but how can I confirm this??
Thank you.
I have followed your tips but I did not get positive results, I still have the same problem.
The dendrimer I use is not a solvent, I solvate both of dendrimer and DNA in water. I simulated the dendrimer alone in water and there was no problem everything went will, but the segmentation fault problem appears when I want to equilibrate the DNA. I downloaded the pdb file of the all atom DNA structure from the pdb.org and used the python script (martinize-dna.py) to convert the all atom to coarse-grained. I have a doubt that the problem is in the structure!! but how can I confirm this??
Thank you.
Please Log in or Create an account to join the conversation.
- jaakko
-
- Offline
- Admin
Less
More
- Posts: 70
9 years 2 months ago #4455
by jaakko
Replied by jaakko on topic Segmentation fault
Hi Doa,
I'm not sure if this is a separate problem from the other message you posted but the same reply applies here. Check that you're using an mdp file meant for DNA. Also using -rdd 1.8 should help keep the DNA strands stable.
- Jaakko
I'm not sure if this is a separate problem from the other message you posted but the same reply applies here. Check that you're using an mdp file meant for DNA. Also using -rdd 1.8 should help keep the DNA strands stable.
- Jaakko
Please Log in or Create an account to join the conversation.
- doa.hawamdeh
-
- Offline
- Fresh Boarder
Less
More
- Posts: 18
9 years 2 months ago #4457
by doa.hawamdeh
Replied by doa.hawamdeh on topic Segmentation fault
Yes this is the same problem.
Thanks.
Doa
Thanks.
Doa
Please Log in or Create an account to join the conversation.
Time to create page: 0.113 seconds