- Posts: 272
Problems with Martinize script
- farrotti
-
Topic Author
- Visitor
11 years 3 months ago #1389
by farrotti
Problems with Martinize script was created by farrotti
Hi developers,
I have some questions about your script "martinize".
At the beginning I performed the "Protein in water" tutorial, in which is included a martinize script.
After that I used this file (with the elnedyn option) on the protein I'm analyzing to obtain the CG structure and topology for a classic 300K MD simulation with no problems. In this case I start the script with the following command:Recently I find on your website the most recent version of martinize script (2.2) and I used it to obtain a new topology file for my protein with the following command:The topology file I obtained was different respect to the first one in the definition of some beads type (C5 in the first case, N0 in the last) and in the "bond", "constraints" and "Backbone angles" sections. On the contrary if I don't use the elnedyn option the files are the same.
So my first question is: why I obtain two different topology files? Is the first martinize script based on a different Martini FF or elnedyn version?
Moreover I used this last topology file (the one of martinize2.2 with elnedyn) as input for my MD simulation. I performed the MD with the same files and procedure as in the first one, but I obtain LINCS warnings on histidine sidechain beads and the simulation crashes after few steps. Are these changes on the topology file the cause of these errors? Which is the "correct" topology?
I could not attach the topology files.
Thank you in advance,
Andrea.
I have some questions about your script "martinize".
At the beginning I performed the "Protein in water" tutorial, in which is included a martinize script.
After that I used this file (with the elnedyn option) on the protein I'm analyzing to obtain the CG structure and topology for a classic 300K MD simulation with no problems. In this case I start the script with the following command:
./martinize.py -f ../2shp-w_da_em2.pdb -o 2shp.top -x 2shp.pdb -ss ../../2shp-ss.dat -p backbone -elnedyn -ef 500 -eu 0.9./martinize.py -f ../2shp-w_da_em2.pdb -o 2shp.top -x 2shp.pdb -ss ../../2shp-ss.dat -p backbone -ff elnedyn -ef 500 -eu 0.9So my first question is: why I obtain two different topology files? Is the first martinize script based on a different Martini FF or elnedyn version?
Moreover I used this last topology file (the one of martinize2.2 with elnedyn) as input for my MD simulation. I performed the MD with the same files and procedure as in the first one, but I obtain LINCS warnings on histidine sidechain beads and the simulation crashes after few steps. Are these changes on the topology file the cause of these errors? Which is the "correct" topology?
I could not attach the topology files.
Thank you in advance,
Andrea.
Please Log in or Create an account to join the conversation.
- djurre
-
- Offline
- Admin
Less
More
11 years 3 months ago #1391
by djurre
Replied by djurre on topic Problems with Martinize script
Hi Andrea,
In earlier versions of the script there have been some bugs. (see also the bugfixes mentioned at the download page). The current version (2.2) gives, to the best of my knowledge, topologies corresponding to the Elnedyn forcefield as published, updated with a few changes that the script mentions (Backbone-Backbone bonds are constraints in stead of high force constant bonds. Trp has an extra constrain added to the sidechain. The Backbone sidechain bonds with high force constants are replaced by constraints except for Trp and His.)
This gives stable simulations for most cases, but also in my experience not always. A few things might improve the stability:
- Definitely set the mass of S-type beads at 72amu, not 45 or 54.
- Make sure there is no tension in your atomistic structure. You could do this by running a few (~20) steps EM with an atomistic FF. But it seems you already did this.
-In my experience it sometimes helps to change the constraints between BB beads to bonds with a high force constant (150,000kJ/mol/nm^2).
The first and third suggestions will not change the ensemble average of you simulations, only the fluctuations. And even that not to much. The second suggestion might change more, but that should be positive since simulating a too tense structure is probably not what you want.
In earlier versions of the script there have been some bugs. (see also the bugfixes mentioned at the download page). The current version (2.2) gives, to the best of my knowledge, topologies corresponding to the Elnedyn forcefield as published, updated with a few changes that the script mentions (Backbone-Backbone bonds are constraints in stead of high force constant bonds. Trp has an extra constrain added to the sidechain. The Backbone sidechain bonds with high force constants are replaced by constraints except for Trp and His.)
This gives stable simulations for most cases, but also in my experience not always. A few things might improve the stability:
- Definitely set the mass of S-type beads at 72amu, not 45 or 54.
- Make sure there is no tension in your atomistic structure. You could do this by running a few (~20) steps EM with an atomistic FF. But it seems you already did this.
-In my experience it sometimes helps to change the constraints between BB beads to bonds with a high force constant (150,000kJ/mol/nm^2).
The first and third suggestions will not change the ensemble average of you simulations, only the fluctuations. And even that not to much. The second suggestion might change more, but that should be positive since simulating a too tense structure is probably not what you want.
Please Log in or Create an account to join the conversation.
- farrotti
-
Topic Author
- Visitor
11 years 3 months ago #1393
by farrotti
Replied by farrotti on topic Problems with Martinize script
Thank you for the suggestions. I'll try these changes on my MDs.
Andrea
Andrea
Please Log in or Create an account to join the conversation.
- junfeng
-
- Offline
- Fresh Boarder
Less
More
- Posts: 13
11 years 2 months ago #1404
by junfeng
Replied by junfeng on topic Problems with Martinize script
I have the same issue with LINCS warning and crashing in MD run with the ELNEDYN model. I find that if the time step is reduced to 10fs instead of 20, the problem vanishes.
Hope it helps
But please let me if the methods djurre suggested could solve the problem.
Hope it helps
But please let me if the methods djurre suggested could solve the problem.
Please Log in or Create an account to join the conversation.
- xavier
-
- Offline
- Admin
Less
More
- Posts: 416
11 years 1 month ago #1437
by xavier
Replied by xavier on topic Problems with Martinize script
Yes the solution proposed by Djurre should fix the problem. Did it?
junfeng wrote: I have the same issue with LINCS warning and crashing in MD run with the ELNEDYN model. I find that if the time step is reduced to 10fs instead of 20, the problem vanishes.
Hope it helps
But please let me if the methods djurre suggested could solve the problem.
Please Log in or Create an account to join the conversation.
- junfeng
-
- Offline
- Fresh Boarder
Less
More
- Posts: 13
11 years 1 month ago #1448
by junfeng
Replied by junfeng on topic Problems with Martinize script
I tried the first and second suggestion and they did not work. Still got LINCS ERRORS. I do not know if the third suggestion could solve the problem.
Now I just run 10fs step with ELNEDYN model.
Now I just run 10fs step with ELNEDYN model.
Please Log in or Create an account to join the conversation.
- xavier
-
- Offline
- Admin
Less
More
- Posts: 416
11 years 1 month ago #1453
by xavier
Replied by xavier on topic Problems with Martinize script
If the change of masse does not help then you can switch the constrains on the backbone-backbone bonds to regular bonds with a 500,000.00 force constant. This is the original set up and although switching to constrains should be fine it results in instability in some cases ...
You could then use a 20 fs time step.
You could then use a 20 fs time step.
junfeng wrote: I tried the first and second suggestion and they did not work. Still got LINCS ERRORS. I do not know if the third suggestion could solve the problem.
Now I just run 10fs step with ELNEDYN model.
Please Log in or Create an account to join the conversation.
Time to create page: 0.098 seconds