- Posts: 6
Reverse transformation from CG to AA
- Codename
- Topic Author
- Offline
- Fresh Boarder
I'm doing some AAMD simulations with a system composed of two big proteins and I tried to do some CGMD simulations too in order to see if it's possible to use a combination of CG/AA MD in order to explore a larger time domain with the resources I have.
I have successfully realized a CG version of my system thanks to the detailed tutorial available here, but now I'm experiencing some problems in the reverse transformation step.
Every time I try to obtain an atomistic .gro from the last frame of my CG simulation, the c_cg2fg tool gives me a segmentation fault error.
At the beginning I used the .gro of the box with the CG water (W) and the tool didn't give me this error, but the coordinates of the protein were always 0.000; then I tried with the box containing only the two proteins (obtained through g_trjconv) and c_cg2fg started to give me this error.
I tried also to convert the proteins individually with a .top including directly the infos about the molecule (usually I work with a topology file with #include to the .itp files of the proteins, but I read on the forum that this could lead to some mistakes during the processing of the topology) but this didn't make any difference.
I'm quite sure about the quality of the topology files of both CG and AA structures (I made several MDs with this topologies without problems), but although this I'm looking for some differences in them at the moment.
Could you please suggest me what I may do in order to solve this problem?
By the way, for the AA model I'm using the ffG43a1 gromos forcefield.
Best Regards,
Paolo Conflitti
Please Log in or Create an account to join the conversation.
- Codename
- Topic Author
- Offline
- Fresh Boarder
- Posts: 6
Nevertheless now I'm facing another problem that it should be simple to solve, but it's strange: the fg2cg tool doesn't recognize the COO- term of the C-term (an alanine residue) of one of my protein. When I compute the Fg pre-structure at the two oxygen atoms coordinates 0.000 are given.
I'm trying to solve this problem with the help of other softwares (specifically, Swiss-PDBViewer) that allow me to manipulate the protein, but I thought that this issue could be of some interest for you.
Paolo Conflitti
Please Log in or Create an account to join the conversation.
- xavier
- Offline
- Admin
- Posts: 416
The problem you experience with the COO- should not happen!
Are you sure that the topologies are correctly defined?
I would suggest you try on a small system for which things would
be easy to do. Water would be easy :))
XAvier.
Codename wrote: I managed to partially solve the problem with the reverse transformation. The modified version of gromacs was not installed properly, so I made a complete new installation that solved the Segmentation fault error.
Nevertheless now I'm facing another problem that it should be simple to solve, but it's strange: the fg2cg tool doesn't recognize the COO- term of the C-term (an alanine residue) of one of my protein. When I compute the Fg pre-structure at the two oxygen atoms coordinates 0.000 are given.
I'm trying to solve this problem with the help of other softwares (specifically, Swiss-PDBViewer) that allow me to manipulate the protein, but I thought that this issue could be of some interest for you.
Paolo Conflitti
Please Log in or Create an account to join the conversation.
- Codename
- Topic Author
- Offline
- Fresh Boarder
- Posts: 6
thanks for the reply, indeed there was a problem with the topology file, the mapping section was almost completely wrong because of a shift in the numbering of the atoms. Now the topology seems to be correct, but I experienced a strange problem.
I'm using two proteins with three histidines protonated on both NE and ND (HISH), but when I try to do the reverse transformation coordinates 0.000 are given to the HD1 hydrogen (the hydrogen of the ND).
While I was trying to found the source of this problem I founded that the mapping section of the FF43A2 (with mapping) seems to be exchanged for the HISH and HISA residue types.
I don't know if this problem is limited to my circumstances or it is irrelevant because it concerns a hydrogen and the residue type takes account for it, but I thought that I had to report it.
Paolo Conflitti
Please Log in or Create an account to join the conversation.
- xavier
- Offline
- Admin
- Posts: 416
Thanks for reporting the problem. It is important to do so and we always appreciate it.
To summarize: you found that the mapping of HISH is not properly done in the current version of the FF topology. It is missing the definition of an H (HD1). You modified it and then you could get the proper description reverse mapping?
Could you post the modified HISH here that we would include in the next release of the force field or people could actually copy it is necessary?
Thanks again for reporting,
XAvier.
Codename wrote: Dear Xavier,
thanks for the reply, indeed there was a problem with the topology file, the mapping section was almost completely wrong because of a shift in the numbering of the atoms. Now the topology seems to be correct, but I experienced a strange problem.
I'm using two proteins with three histidines protonated on both NE and ND (HISH), but when I try to do the reverse transformation coordinates 0.000 are given to the HD1 hydrogen (the hydrogen of the ND).
While I was trying to found the source of this problem I founded that the mapping section of the FF43A2 (with mapping) seems to be exchanged for the HISH and HISA residue types.
I don't know if this problem is limited to my circumstances or it is irrelevant because it concerns a hydrogen and the residue type takes account for it, but I thought that I had to report it.
Paolo Conflitti
Please Log in or Create an account to join the conversation.
- Codename
- Topic Author
- Offline
- Fresh Boarder
- Posts: 6
here it is the modified version of mapping sections of HISA and HISH. They seemed to be only exchanged, and after the correction I managed to get the proper reverse mapping without problems.
HISA
[ atoms ]
N N -0.28000 0
H H 0.28000 0
CA CH1 0.00000 1
CB CH2 0.00000 1
CG C 0.00000 2
ND1 NR 0.00000 2
HD1 H 0.19000 2
CD2 CR1 0.13000 2
CE1 CR1 0.26000 2
NE2 NR -0.58000 2
C C 0.380 3
O O -0.380 3
[ bonds ]
N H gb_2
N CA gb_20
CA C gb_26
C O gb_4
C +N gb_9
CA CB gb_26
CB CG gb_26
CG ND1 gb_9
CG CD2 gb_9
ND1 HD1 gb_2
ND1 CE1 gb_9
CD2 NE2 gb_9
CE1 NE2 gb_9
[ exclusions ]
; ai aj
CB HD1
CB CE1
CB NE2
HD1 CD2
HD1 NE2
[ angles ]
; ai aj ak gromos type
-C N H ga_31
H N CA ga_17
-C N CA ga_30
N CA C ga_12
CA C +N ga_18
CA C O ga_29
O C +N ga_32
N CA CB ga_12
C CA CB ga_12
CA CB CG ga_14
CB CG ND1 ga_36
CB CG CD2 ga_36
ND1 CG CD2 ga_6
CG ND1 HD1 ga_35
CG ND1 CE1 ga_6
HD1 ND1 CE1 ga_35
CG CD2 NE2 ga_6
ND1 CE1 NE2 ga_6
CD2 NE2 CE1 ga_6
[ impropers ]
; ai aj ak al gromos type
N -C CA H gi_1
C CA +N O gi_1
CA N C CB gi_2
CG ND1 CD2 CB gi_1
CD2 CG ND1 CE1 gi_1
ND1 CG CD2 NE2 gi_1
CG ND1 CE1 NE2 gi_1
CG CD2 NE2 CE1 gi_1
CD2 NE2 CE1 ND1 gi_1
ND1 CG CE1 HD1 gi_1
[ dihedrals ]
; ai aj ak al gromos type
-CA -C N CA gd_4
-C N CA C gd_19
N CA C +N gd_20
N CA CB CG gd_17
CA CB CG ND1 gd_20
[ mapping ]
; bead atom
1 N
1 H
1 H1
1 H2
1 H3
1 CA
1 C
1 O
1 O1
1 O2
2 CB
2 CG
3 CD2
3 HD2
3 NE2
4 ND1
4 HD1
4 CE1
4 HE1
HISH
[ atoms ]
N N -0.28000 0
H H 0.28000 0
CA CH1 0.00000 1
CB CH2 0.00000 1
CG C -0.05000 2
ND1 NR 0.38000 2
HD1 H 0.30000 2
CD2 CR1 0.00000 2
CE1 CR1 -0.24000 2
NE2 NR 0.31000 2
HE2 H 0.30000 2
C C 0.380 3
O O -0.380 3
[ bonds ]
N H gb_2
N CA gb_20
CA C gb_26
C O gb_4
C +N gb_9
CA CB gb_26
CB CG gb_26
CG ND1 gb_9
CG CD2 gb_9
ND1 HD1 gb_2
ND1 CE1 gb_9
CD2 NE2 gb_9
CE1 NE2 gb_9
NE2 HE2 gb_2
[ exclusions ]
; ai aj
CB HD1
CB CE1
CB NE2
CG HE2
ND1 HE2
HD1 CD2
HD1 NE2
[ angles ]
; ai aj ak gromos type
-C N H ga_31
H N CA ga_17
-C N CA ga_30
N CA C ga_12
CA C +N ga_18
CA C O ga_29
O C +N ga_32
N CA CB ga_12
C CA CB ga_12
CA CB CG ga_14
CB CG ND1 ga_36
CB CG CD2 ga_36
ND1 CG CD2 ga_6
CG ND1 HD1 ga_35
CG ND1 CE1 ga_6
HD1 ND1 CE1 ga_35
CG CD2 NE2 ga_6
ND1 CE1 NE2 ga_6
CD2 NE2 CE1 ga_6
CD2 NE2 HE2 ga_35
CE1 NE2 HE2 ga_35
[ impropers ]
; ai aj ak al gromos type
N -C CA H gi_1
C CA +N O gi_1
CA N C CB gi_2
CG ND1 CD2 CB gi_1
CD2 CG ND1 CE1 gi_1
ND1 CG CD2 NE2 gi_1
CG ND1 CE1 NE2 gi_1
CG CD2 NE2 CE1 gi_1
CD2 NE2 CE1 ND1 gi_1
ND1 CG CE1 HD1 gi_1
NE2 CD2 CE1 HE2 gi_1
[ dihedrals ]
; ai aj ak al gromos type
-CA -C N CA gd_4
-C N CA C gd_19
N CA C +N gd_20
N CA CB CG gd_17
CA CB CG ND1 gd_20
[ mapping ]
; bead atom
1 N
1 H
1 H1
1 H2
1 H3
1 CA
1 C
1 O
1 O1
1 O2
2 CB
2 CG
3 CD2
3 HD2
3 NE2
3 HE2
4 ND1
4 HD1
4 CE1
4 HE1
I'm glad that I could be of help, thanks to you for all the work you're doing.
Paolo Conflitti
Please Log in or Create an account to join the conversation.
- Codename
- Topic Author
- Offline
- Fresh Boarder
- Posts: 6
I'm experiencing a problem with the SA. After the reverse transformation from CG to AA I had to reintroduce some Na+ atoms in order to neutralize the system, and then I tried to do the SA using the mdp included in the tutorial and modifying it for the 1000 step, 5000 step and 60000 step SA.
For reintroducing the ions I used the genion tool, while in order to obtain the reverse transformation without the problem of mapping the Na+ atom I simply substituted the ions with W through manual editing of the cg.gro file.
The 1000 and 5000 step went well, while the 40000 and 60000 step crashed several times, giving back the error "Cannot put the x atom on the grid, NaN" or "The charge went out of the box". These errors seem to be caused by several constraint deviations that lead to lincs warning, but the atoms that give these deviations are FG_W atoms only.
I experienced this error only for the 40000 and the 60000 step SA, but when I modified the time of annealing to less of the half of the original value described in the tutorial I managed to complete them without these error.
I tried several attempts in order to solve this problem, but till now only the one above mentioned seems to work, even if in a way that isn't really useful to me. I tried also to modify the table extension size because gromacs found 1-4 interactions at a distance far greater than 1.2 nm, around 1.6 nm, but this was useless too.
Now I'm worried that maybe there's something wrong with my system, even if it doesn't seem to be exploding, so I decided to look for any help here. I'm sorry that I'm still bothering you, but I really don't know what to do.
Thanks in advance for any help you could give me,
Paolo Conflitti
Please Log in or Create an account to join the conversation.
- andrzej
- Visitor
Numbers of things you can do to remove crashes:
-decrease cap_force
-decrease cap_a
-decrease time step
-increase values for lincs: lincs_order, lincs_iter
Andrzej
Please Log in or Create an account to join the conversation.
- Codename
- Topic Author
- Offline
- Fresh Boarder
- Posts: 6
Please Log in or Create an account to join the conversation.