CHARMM Development Project Forums User Discussion & Questions Energy terms, Constraints, Restraints, and Solvation CMAP correction, !NCRTERM definitions for multiple solutes

Hi
I am working on simulating lipid bilayers with previtamin D. I have generated a CMAP for the open-ring portion of the previtamin D molecule and I want to apply it to my system.

The membrane contains a total of eight Pre molecules and I am unsure how to get the CMAP to apply to all the molecules and not just RESID 1. My assumption is that the !NCRTERM portion of step5_assembly.psf requires one line for each dihedral that will be modified by the CMAP, with the atom types identified in my ligand rtf file. If I have eight ligands then I should have eight lines.

The CHARMM output states that the additional cross-terms that I have defined do not have an associated CMAP. My confusion stems from the fact that the atoms I selected in the !NCRTERM block do not correspond to the same types shown in output. I used the atom indices from the psf file to select the atoms and in that file they have the correct type.

Ultimately what I am looking to find out is this: do I need a new cross-term for each separate ligand or will CHARMM recognize one cross-term and apply it to the other ligands for me? (Handling multiple cross-terms is not discussed in the prm portion of the user manual)


I have attached four files: the output from my attempted run, the coordinates and cross-term blocks from step5_assembly.psf (full file too large), and the rtf and prm file for my previtamin D ligand.

Also, below is a highlight showing how the atom labels are not corresponding with what I am expecting.

From cross-terms block of psf:
8 !NCRTERM: cross-terms
67 54 51 50 51 50 48 1
139 126 123 122 123 122 120 73
211 198 195 194 195 194 192 145
283 270 267 266 267 266 264 217
355 342 339 338 339 338 336 289
427 414 411 410 411 410 408 361
499 486 483 482 483 482 480 433
571 558 555 554 555 554 552 505

From atoms block psf:
139 HETA 2 LIG C20 CG321 -0.201000 12.0110 0 0.00000 -0.301140E-02
126 HETA 2 LIG C19 CG2DC1 0.190000E-01 12.0110 0 0.00000 -0.301140E-02
123 HETA 2 LIG C17 CG2DC2 -0.149000 12.0110 0 0.00000 -0.301140E-02
122 HETA 2 LIG C15 CG2DC2 -0.148000 12.0110 0 0.00000 -0.301140E-02
120 HETA 2 LIG C9 CG2DC1 0.350000E-01 12.0110 0 0.00000 -0.301140E-02
73 HETA 2 LIG C3 CG3RC1 -0.132000 12.0110 0 0.00000 -0.301140E-02


From output:
: No cross-term map for 2 ( CG321 HAL2 CTL2 HGA3 ) ( HGA1 HGA3 HGA1 HGA1 )
: No cross-term map for 3 ( HGA2 CG2DC1 HAL2 HGA2 ) ( HGA3 HGA2 HGA3 HGA2 )
: No cross-term map for 4 ( HGA3 HGA3 HGA3 PL ) ( HAL2 PL HAL2 CG321 )
: No cross-term map for 5 ( HGA2 HGA3 HGA2 CTL2 ) ( CTL3 CTL2 CTL3 HGA2 )
: No cross-term map for 6 ( CG321 HAL2 HGA3 HGA2 ) ( CG321 HGA2 HGA3 HGA3 )
: No cross-term map for 7 ( HGA4 CTL2 CTL2 CG321 ) ( CG3C52 CG321 CG3C52 CG3C51 )
: No cross-term map for 8 ( HGA3 HGA3 HAL2 CG2DC2 ) ( CG331 CG2DC2 CG331 HGA3 )
: A TOTAL OF 7 MISSING PARAMETERS

If the atom parameter types are are the same in each ligand, only a single CMAP definition should be needed; there's only one CMAP definition for the backbone heavy atoms of amino acid polymers.

There does seem to be an issue, but I cannot tell exactly what it might be from the data included so far. The only thing I noticed was this message--

PARRDR> WARNING: ATOMS IN PHI1 (CMAP) STOP DONT EXIST
PARMIO> NONBOND, HBOND lists and IMAGE atoms cleared.

and the fact that the .prm file was terminated incorrectly with a STOP statement instead of an END statement. Fix that and try again.

Thank you for the quick response Rick. So if I am understanding correctly, I only need the first definition for the cross-terms block in the psf file and CHARMM will apply that to the other 7 Pre molecules.

I changed the terminal command in the prm file from stop to end as you suggested and the warning disappeared. Thanks for catching that!!

I believe each instance should be listed in the PSF, but only a single definition and data table should be needed in the PARAM file.

Did the missing cross-term map error messages go away?

I am still getting the missing cross-term map error, only for some reason it is declaring a different set of atom types than in the original error message, even though the atom selections were the same:

8 !NCRTERM: cross-terms
67 54 51 50 51 50 48 1
139 126 123 122 123 122 120 73
211 198 195 194 195 194 192 145
283 270 267 266 267 266 264 217
355 342 339 338 339 338 336 289
427 414 411 410 411 410 408 361
499 486 483 482 483 482 480 433
571 558 555 554 555 554 552 505


: No cross-term map for 2 ( CG321 CG311 HGA3 CG2DC2 ) ( CG2DC2 CG2DC2 CG2DC2 CG2DC1 )
: No cross-term map for 3 ( HGA2 HGA3 HGA3 CG2DC2 ) ( CG2DC2 CG2DC2 CG2DC2 CG2DC1 )
: No cross-term map for 4 ( HGA3 CG3C52 HGA2 CG2DC2 ) ( CG2DC2 CG2DC2 CG2DC2 CG2DC1 )
: No cross-term map for 5 ( HGA2 HGA4 CG321 CG2DC2 ) ( CG2DC2 CG2DC2 CG2DC2 CG2DC1 )
: No cross-term map for 6 ( CG321 CG3RC1 HGA2 CG2DC2 ) ( CG2DC2 CG2DC2 CG2DC2 CG2DC1 )
: No cross-term map for 7 ( HGA4 HGA1 HGA2 CG2DC2 ) ( CG2DC2 CG2DC2 CG2DC2 CG2DC1 )
: No cross-term map for 8 ( HGA3 HGA2 HGA3 CG2DC2 ) ( CG2DC2 CG2DC2 CG2DC2 CG2DC1 )
: A TOTAL OF 7 MISSING PARAMETERS

It's not something I've seen before, and it's not clear what the issue is. A bug? A user input error?

I looked at cmap_prm.txt and it wasn't clear to me if these were two adjacent dihedrals, how they overlapped, and if the atom sequences were in the correct order.

Try a simpler test case, with only the ligand; start with just one, then two, and three, and see where it breaks down.

Is it possible for me to use the .prm and .rtf files from my original system, which already contain the CMAP stub and the dihedral definition, and load them manually during Step 1 of the Membrane Builder instead of generating new files? My hope is that CHARMM will automatically include the cross-terms in the Step5_assembly.psf file and use the correct atom typing. I would just cut that block and paste it into the psf for the original system.

You may need to ask the charmm-gui people about supplying RTF and PARAM for your ligand for the Membrane Builder, using their Contact page on that site.

Manual modification of a PSF is usually a bad idea.

Yes this is where I was having trouble. I had underestimated the sensitivity of the psf file when I was adding my cross-terms. The alignment of the numbers is apparently significant and I was not aware of this. I had been trying to keep the same quantity of delimited spaces between each column instead of trying to keep the numbers in the columns aligned.

I attached an image of the correct alignment that I am now using.

This was not very apparent in my OP as the forum reformatted the text in the !NCRTERM block, though it is visible at the bottom of psf.txt file... but this made all the difference. CHARMM no longer crashes with the MISSING PARAMETERS error.

Thank you @rmv for your help.