- Posts: 10
seq2itp.pl vs martinize.py
- jantunes
-
Topic Author
- Offline
- Fresh Boarder
Less
More
12 years 6 months ago #825
by jantunes
seq2itp.pl vs martinize.py was created by jantunes
Hey,
1.
I found a difference between the seq2itp.pl and martinize.py scripts. The first determines the angle/force of 3 residues from the secondary structure of the middle one (H-H-C is 96o/700 and H-C-C is 127o/25). The second determines the angle/force from the secondary structure with the lowest force like in Martini: Extension to proteins (H-H-C and H-C-C are 127o/25). Is that made with propose? What should I use?
2.
I also found a (small) error, I think, in one echo of seq2itp.pl, that say "Force constants follow those presented in Monticelli et al. (2008) with the exception of coil angle force constant that is slightly increased to improve the stability of the model in some cases" shouldn't it be "of coil bond force constant", that changed from 200 (Monticelli et.al 2008) to 400. right?
3.
Martinize.py and atom2cg.awk give (kinda small) different positions to beads, but I think those should be because different precisions in calculations, no?
3.
Finally my last problem (and kinda offtopic) is that I can't change one DMPC lipid in fg to cg with both scripts, and I don't like that much about using martini_v2.0_lipids.itp becouse it is very reductionist about beads. Like 4xC bead be equal to 3xC bead, and DMPC with less one carbon gets one less bead than DPPC. Not to talk that I don't see how to make a gro file of dmpc/dlpc. Any suggestion/help?
Thanks :)
Jorge
1.
I found a difference between the seq2itp.pl and martinize.py scripts. The first determines the angle/force of 3 residues from the secondary structure of the middle one (H-H-C is 96o/700 and H-C-C is 127o/25). The second determines the angle/force from the secondary structure with the lowest force like in Martini: Extension to proteins (H-H-C and H-C-C are 127o/25). Is that made with propose? What should I use?
2.
I also found a (small) error, I think, in one echo of seq2itp.pl, that say "Force constants follow those presented in Monticelli et al. (2008) with the exception of coil angle force constant that is slightly increased to improve the stability of the model in some cases" shouldn't it be "of coil bond force constant", that changed from 200 (Monticelli et.al 2008) to 400. right?
3.
Martinize.py and atom2cg.awk give (kinda small) different positions to beads, but I think those should be because different precisions in calculations, no?
3.
Finally my last problem (and kinda offtopic) is that I can't change one DMPC lipid in fg to cg with both scripts, and I don't like that much about using martini_v2.0_lipids.itp becouse it is very reductionist about beads. Like 4xC bead be equal to 3xC bead, and DMPC with less one carbon gets one less bead than DPPC. Not to talk that I don't see how to make a gro file of dmpc/dlpc. Any suggestion/help?
Thanks :)
Jorge
Please Log in or Create an account to join the conversation.
- djurre
-
- Offline
- Admin
Less
More
- Posts: 272
12 years 6 months ago #827
by djurre
Replied by djurre on topic seq2itp.pl vs martinize.py
Hey Jorge,
1: You are completely right. The Martini protein article (Monticelli, 2008) states: "When a backbone bonded parameter (either a bond or an angle) involves beads with more than one type of secondary structure, the weaker force constant is used." It turned out that the seq2itp.pl script did not do that, as you already mentioned. In the new script we have now implemented this correctly (I hope...).
However it makes a difference only in specific cases and for the overall dynamics of the protein I think it will make no difference what so ever.
2: You are completely right again.
3: The atom2CG.awk script places the CG bead on the positions of certain AA-atoms (e.g.: BB at the position of the CA). This is no problem, because it is a starting structure only and will relax it's optimal positions quickly. It will matter however if you use the starting structure do determine bonds for an elastic network!
The new Martinize.py scripts puts the beads at the COM positions (for Martini) and determines elastic bonds from there. If you use it to generate an Elnedyn topology, it puts the backbone beads on the Ca position.
4: Both scripts only handle proteins (for now). Once you have a protein topology/structure you can combine it with your bilayer. How to prepare a gro file for a bilayer is described in the tutorial on this webpage (Tutorial 1a,1b,1d). The fact that the lipid models are to reductionistic is inherent to a CG model: you map 4 atoms to 1 bead so you can not differentiate the tail length per atom.
1: You are completely right. The Martini protein article (Monticelli, 2008) states: "When a backbone bonded parameter (either a bond or an angle) involves beads with more than one type of secondary structure, the weaker force constant is used." It turned out that the seq2itp.pl script did not do that, as you already mentioned. In the new script we have now implemented this correctly (I hope...).
However it makes a difference only in specific cases and for the overall dynamics of the protein I think it will make no difference what so ever.
2: You are completely right again.
3: The atom2CG.awk script places the CG bead on the positions of certain AA-atoms (e.g.: BB at the position of the CA). This is no problem, because it is a starting structure only and will relax it's optimal positions quickly. It will matter however if you use the starting structure do determine bonds for an elastic network!
The new Martinize.py scripts puts the beads at the COM positions (for Martini) and determines elastic bonds from there. If you use it to generate an Elnedyn topology, it puts the backbone beads on the Ca position.
4: Both scripts only handle proteins (for now). Once you have a protein topology/structure you can combine it with your bilayer. How to prepare a gro file for a bilayer is described in the tutorial on this webpage (Tutorial 1a,1b,1d). The fact that the lipid models are to reductionistic is inherent to a CG model: you map 4 atoms to 1 bead so you can not differentiate the tail length per atom.
Please Log in or Create an account to join the conversation.
- jantunes
-
Topic Author
- Offline
- Fresh Boarder
Less
More
- Posts: 10
12 years 6 months ago #828
by jantunes
Replied by jantunes on topic seq2itp.pl vs martinize.py
Hey,
1.
So martinize.py it is, gives half the work also :)
4.
Meh, I didn't checked tutorial 1.d. I have conscience about the reductionism of Martini, but lipids in martini_v2.0_lipids.itp are too much. For exemple, In "Martini: CG model for biomolecular simulations" there are 5 types of apolar beads, so maybe to make DMPC I could change 2 C1 beads for 2 C2 beads (C1-C1-C2-C2) in DMPC topology, and use a DPPC gro? But dmpc.gro beads positions won't be a problem?
5. (new)
I was digging even more deep in the itp's and I found that diedrals angles just give for 4 residues in helix structure, and not for 4 in extended structure, and the angle is -120 (in itp) and not 60 (in Monticelli et. al 2008). Is there some place stating this changes, a paper or?
Cumps,
Jorge
1.
So martinize.py it is, gives half the work also :)
4.
Meh, I didn't checked tutorial 1.d. I have conscience about the reductionism of Martini, but lipids in martini_v2.0_lipids.itp are too much. For exemple, In "Martini: CG model for biomolecular simulations" there are 5 types of apolar beads, so maybe to make DMPC I could change 2 C1 beads for 2 C2 beads (C1-C1-C2-C2) in DMPC topology, and use a DPPC gro? But dmpc.gro beads positions won't be a problem?
5. (new)
I was digging even more deep in the itp's and I found that diedrals angles just give for 4 residues in helix structure, and not for 4 in extended structure, and the angle is -120 (in itp) and not 60 (in Monticelli et. al 2008). Is there some place stating this changes, a paper or?
Cumps,
Jorge
Please Log in or Create an account to join the conversation.
- djurre
-
- Offline
- Admin
Less
More
- Posts: 272
12 years 5 months ago #839
by djurre
Replied by djurre on topic seq2itp.pl vs martinize.py
Hey Jorge,
The answer is a bit late but:
4. If I understand it correctly, you want to change the beadtype in the itp and use the 'old' gro. That you can do. The positions in the gro probably wont be problem. However, keep in mind you basically create a new forcefield. This forcefield will behave differently then Martini (possibly better), which you will have to investigate and describe (area per lipids, phase transition temperature, etc).
5. We (mostly) don't use the backbone dihedral angles for extended structures any more. It turns out they make the system rather unstable. Instead of that we use local elastic networks. If you want dihedrals anyway use the option "-ed" in martinize.py. This has not yet been described anyway (but probably will be somethime soon). The 60/-120 option is due to how dihedral angles are defined in Gromacs ('maximum of potential' instead of 'maximum of angle').
The answer is a bit late but:
4. If I understand it correctly, you want to change the beadtype in the itp and use the 'old' gro. That you can do. The positions in the gro probably wont be problem. However, keep in mind you basically create a new forcefield. This forcefield will behave differently then Martini (possibly better), which you will have to investigate and describe (area per lipids, phase transition temperature, etc).
5. We (mostly) don't use the backbone dihedral angles for extended structures any more. It turns out they make the system rather unstable. Instead of that we use local elastic networks. If you want dihedrals anyway use the option "-ed" in martinize.py. This has not yet been described anyway (but probably will be somethime soon). The 60/-120 option is due to how dihedral angles are defined in Gromacs ('maximum of potential' instead of 'maximum of angle').
Please Log in or Create an account to join the conversation.
- pimfrederix
-
- Offline
- Fresh Boarder
Less
More
- Posts: 6
11 years 7 months ago #1144
by pimfrederix
Replied by pimfrederix on topic seq2itp.pl vs martinize.py
Good stuff.
One more thing: using martinize.py with peptides: "If polarizable water is used in combination with proteins or peptides, all AC1 and AC2 beads have to be replaced by normal C1 and C2 beads. AC1 and AC2 are obsolete in polarizable Martini."
Can there be an option in martinize.py to do this automatically rather than post-processing the output?
Thanks,
Pim
One more thing: using martinize.py with peptides: "If polarizable water is used in combination with proteins or peptides, all AC1 and AC2 beads have to be replaced by normal C1 and C2 beads. AC1 and AC2 are obsolete in polarizable Martini."
Can there be an option in martinize.py to do this automatically rather than post-processing the output?
Thanks,
Pim
Please Log in or Create an account to join the conversation.
Time to create page: 0.095 seconds