CHARMM Development Project Forums User Discussion & Questions Parameter Set Discussion Trifluoromethylated residue

Hi,

I used ParamChem to obtain the parameters of a trifluoromethylated alanine (see mol2 file below). The parameters were generated successfully (see str file below) and the dihedral with high penalty (50.5) was optimized by fitting on a QM energy scan.

I have a question regarding the charges of the side chain atoms. The charges assigned by ParamChem are more similar to those of the trifluoroethanol (RESI TFE from file top_all36_cgenff.rtf) than those of the trifluoroethane (RESI TFET). This suggests that ParamChem chose TFE as starting compound rather than TFET, which is not what I expected. Is there a reason for this choice considering that parameters for TFET have been thoroughly optimized (J. Comp. Chem. 2002, 23:199-213). Should I change the charges to those of TFET (i.e. -0.15 for atoms F0* and ~ 0.38 for C01)?

Thanks,
Sebastien


trifluoromethylated alanine mol2 file:

#Caution:
#Input/Original file: tfet-ala.pdb
#This tfet-ala.mol2 is synthesized from tfet-ala.pdb using Open Babel 2.3.0
#

@MOLECULE
DR_tfet-ala.pdb
25 24 0 0 0
SMALL
GASTEIGER

@ATOM
1 C01 3.4510 -0.9860 -0.4740 C.3 1 ALA1 0.3923
2 F01 3.8470 -0.5260 0.7320 F 1 ALA1 -0.1707
3 F02 3.6690 -2.3170 -0.5420 F 1 ALA1 -0.1707
4 F03 4.1530 -0.3660 -1.4470 F 1 ALA1 -0.1707
5 CAY -2.2970 0.0840 0.7600 C.3 1 ALA1 0.0122
6 HY1 -2.6430 0.1380 -0.2940 H 1 ALA1 0.0322
7 HY2 -2.4640 1.0650 1.2530 H 1 ALA1 0.0322
8 HY3 -2.8890 -0.6910 1.2930 H 1 ALA1 0.0322
9 CY -0.8620 -0.2750 0.8100 C.2 1 ALA1 0.2105
10 OY -0.4860 -1.3080 1.3440 O.2 1 ALA1 -0.2761
11 N -0.0140 0.5850 0.2410 N.am 1 ALA1 -0.3039
12 HN -0.3310 1.4450 -0.1530 H 1 ALA1 0.1494
13 CA 1.4270 0.4210 0.2820 C.3 1 ALA1 0.1075
14 HA 1.7230 0.2110 1.3000 H 1 ALA1 0.0597
15 CB 1.9490 -0.7030 -0.6650 C.3 1 ALA1 0.0696
16 HB1 1.3840 -1.6330 -0.4250 H 1 ALA1 0.0375
17 HB2 1.7360 -0.4390 -1.7250 H 1 ALA1 0.0375
18 C 2.0010 1.7710 -0.0980 C.2 1 ALA1 0.2334
19 O 1.2570 2.7370 -0.2620 O.2 1 ALA1 -0.2738
20 NT 3.3330 1.8770 -0.2420 N.am 1 ALA1 -0.3168
21 HNT 3.8730 1.0370 -0.1610 H 1 ALA1 0.1484
22 CAT 3.9910 3.1200 -0.5670 C.3 1 ALA1 0.0001
23 HT1 3.7840 3.8900 0.2100 H 1 ALA1 0.0426
24 HT2 5.0920 2.9730 -0.6290 H 1 ALA1 0.0426
25 HT3 3.6340 3.5100 -1.5470 H 1 ALA1 0.0426
@BOND
1 17 15 1
2 25 22 1
3 4 1 1
4 15 1 1
5 15 16 1
6 15 13 1
7 24 22 1
8 22 20 1
9 22 23 1
10 3 1 1
11 1 2 1
12 6 5 1
13 19 18 2
14 20 21 1
15 20 18 am
16 12 11 1
17 18 13 1
18 11 13 1
19 11 9 am
20 13 14 1
21 5 9 1
22 5 7 1
23 5 8 1
24 9 10 2


trifluoromethylated alanine str file:

* Toppar stream file generated by
* CHARMM General Force Field (CGenFF) program version 0.9.7 beta
* For use with CGenFF version 2b8
*

read rtf card append
* Topologies generated by
* CHARMM General Force Field (CGenFF) program version 0.9.7 beta
*
36 1

! "penalty" is the highest penalty score of the associated parameters.
! Penalties lower than 10 indicate the analogy is fair; penalties between 10
! and 50 mean some basic validation is recommended; penalties higher than
! 50 indicate poor analogy and mandate extensive validation/optimization.

RESI ALAF 0.000 ! param penalty= 50.500 ; charge penalty= 11.750
GROUP ! CHARGE CH_PENALTY
ATOM C01 CG302 0.325 ! 11.750
ATOM F01 FGA3 -0.140 ! 2.500
ATOM F02 FGA3 -0.140 ! 2.500
ATOM F03 FGA3 -0.140 ! 2.500
ATOM CAY CG331 -0.271 ! 0.000
ATOM HY1 HGA3 0.090 ! 0.000
ATOM HY2 HGA3 0.090 ! 0.000
ATOM HY3 HGA3 0.090 ! 0.000
ATOM CY CG2O1 0.513 ! 0.000
ATOM OY OG2D1 -0.508 ! 0.000
ATOM N NG2S1 -0.441 ! 4.557
ATOM HN HGP1 0.305 ! 0.000
ATOM CA CG311 0.068 ! 9.504
ATOM HA HGA1 0.090 ! 0.600
ATOM CB CG321 -0.088 ! 11.080
ATOM HB1 HGA2 0.090 ! 0.000
ATOM HB2 HGA2 0.090 ! 0.000
ATOM C CG2O1 0.515 ! 4.557
ATOM O OG2D1 -0.518 ! 0.000
ATOM NT NG2S1 -0.490 ! 0.000
ATOM HNT HGP1 0.308 ! 0.000
ATOM CAT CG331 -0.108 ! 0.000
ATOM HT1 HGA3 0.090 ! 0.000
ATOM HT2 HGA3 0.090 ! 0.000
ATOM HT3 HGA3 0.090 ! 0.000

BOND HB2 CB
BOND HT3 CAT
BOND F03 C01
BOND CB C01
BOND CB HB1
BOND CB CA
BOND HT2 CAT
BOND CAT NT
BOND CAT HT1
BOND F02 C01
BOND C01 F01
BOND HY1 CAY
BOND O C
BOND NT HNT
BOND NT C
BOND HN N
BOND C CA
BOND N CA
BOND N CY
BOND CA HA
BOND CAY CY
BOND CAY HY2
BOND CAY HY3
BOND CY OY
IMPR CY CAY N OY
IMPR C CA NT O

END

read param card flex append
* Parameters generated by analogy by
* CHARMM General Force Field (CGenFF) program version 0.9.7 beta
*

! Penalties lower than 10 indicate the analogy is fair; penalties between 10
! and 50 mean some basic validation is recommended; penalties higher than
! 50 indicate poor analogy and mandate extensive validation/optimization.

BONDS

ANGLES
CG302 CG321 CG311 58.35 113.50 11.16 2.56100 ! DR_tfe , from CG311 CG321 CG321, penalty= 12

DIHEDRALS
FGA3 CG302 CG321 CG311 0.2500 3 0.00 ! DR_tfe , from FGA3 CG302 CG321 OG311, penalty= 50.5
CG2O1 CG311 CG321 CG302 0.2000 3 0.00 ! DR_tfe , from CG2O1 CG311 CG321 CG321, penalty= 12
NG2S1 CG311 CG321 CG302 0.2000 3 0.00 ! DR_tfe , from NG2S1 CG311 CG321 CG321, penalty= 12
HGA1 CG311 CG321 CG302 0.1950 3 0.00 ! DR_tfe , from HGA1 CG311 CG321 CG321, penalty= 12

IMPROPERS

END
RETURN

The ParamChem charge assignment is described in K. Vanommeslaeghe, E. P. Raman, A. D. MacKerell Jr., J. Chem. Inf. Model. 2012, 52, 3155-3168.

The two charge sets in the main file are very close; I don't expect significant differences in the properties. Each of the fluorines is 0.01 less negative in TFE, but the overall CF3 group is 0.01 more negative (the slightly larger difference for the C is of little consequence because that atom is buried). The sign of the difference for the overall CF3 group doesn't look very physical, but again, the magnitude is insignificant. The person who did this was probably fine-tuning water interaction energies within a very small margin. If it would be me doing it now, I would keep them the same for the sake of transferability and to avoid overfitting, but at that time (1999), CGenFF wasn't conceived yet and people were less concerned about transferability.

Moving on to the subject of your molecule, this is a case where manual charge assignment would be unambiguous, so you don't need to use the ParamChem charges (though it wouldn't be a huge mistake to do so). For the purpose of transferring charges manually, both charge sets in the main file are probably equally good. TFE has an oxygen in the same position as your C(alpha), and the oxygen is probably a stronger electron withdrawing group, but TFET has no electron withdrawing group at all. At the end of the day, since there is less potential for secondary interactions in the TFET charge optimization, and that molecule was studied extensively in the paper you're citing, I would agree with you in feeling slightly more comfortable transferring from TFET, but yet again, it matters little.