Starting with version 6, YASARA can derive (Y)AMBER force field parameters for unknown molecules fully automatically, allowing to simulate 98% of the structures in the PDB at the touch of a button. Manual intervention is only needed in case of exotic metal ions. The approach behind the AutoSMILES algorithm can be summarized as follows:

• Assignment of pH dependent fractional bond orders and protonation patterns, typing of ring systems by a graph-theoretic approach.
• Identification of known molecules (from the force field definition file or the AMBER Parameter Database using SMILES strings. If no hit is found, proceed to step 3).
• Calculation of semi-empirical AM1 Mulliken point charges[3]. This step involves a geometry optimization with the COSMO solvation model[4] and avoids fatal rearrangements sometimes found when optimizing highly charged molecules like ATP4- in vacuo.
• Assignment of AM1BCC atom- and bond types.
• Application of the 'AM1 Bond Charge Correction' to improve the AM1 charges and make them better represent the electrostatic potential around the molecule - just like RESP charges.
• Further improvement of the AM1BCC charges using the known ideal RESP charges of similar molecule fragments, identified via SMILES strings.
• Assignment of GAFF (General AMBER Force Field) atom types and remaining force field parameters.
• In the end, the newly created parameters are cached for instantaneous availability next time.

Since all this is done automatically, only one step is required in practice: Press <F12> to run the simulation. YASARA also produces a detailed force field parameter assignment report in the console, which you can examine in detail to see what happened.

If a residue contains more atoms than YASARA's QM module can handle , YASARA will try to split it up into smaller pieces that can easily be parameterized independently. While this works well for lipids, where each hydrophobic tail is usually parameterized separately, you may have to give YASARA a hint for other large molecules by following these steps:

• Identify an aliphatic carbon that is as far away as possible from polar atoms, and - if removed - would split your residue in two parts in the soup (i.e. all atoms with lower numbers precede the carbon, all atoms with higher numbers follow it in the soup).
• Mark the carbon, then right-click to activate the context menu and click Split > Atom.
• In the bottom sequence selector, your residue should now have been split in exactly two parts. Now you can run the simulation.

If you nevertheless want to define force field parameters manually , this implies adding at least a residue topology in AMBER PREP format to one of the force field definition files (*.fof) in the yasara/fof subdirectory. The best location is probably gafftopo.fof, since this file is included in all force fields.

Try to follow these steps for AMBER-style force fields:

• Look at the repository of AMBER force field parameters at http://pharmacy.man.ac.uk/amber/. Maybe your molecule has already been parameterized. Also try to Google it. Note that most of the AMBER Parameter Database is already included in YASARA by default.

• Read the introduction to parameter fitting on page 287 of the AMBER 7 manual. (Downloadable from http://amber.scripps.edu).

• Exit YASARA and open the force field definition file in a text editor. You can find the force fields at yasara/fof/ForceFieldName.fof, e.g. yasara/fof/yamber2.fof for the YamberII force field.

• Go to the end of the file and add a topology entry for your residue. The format of these topology entries is also described on the AMBER website. YASARA does not use the bond length, angle and dihedral data, so these columns can be set to zero.

The information needed for every atom is the sequential number (column 1), the atom name in the PDB file (column 2), the force field atom type (column 3) and the point charge on the atom (last column). Just keep the header (including the three DUMMy atoms), and replace the name of the compound (line 1) and the three letter code of your residue (line 3, 'AGS' in the example below). The second line must stay empty.

If your ligand contains planar groups (around resonance or double bonds), you must also add improper dihedral entries (see IMPROPER statements below).

The LOOP statement used by AMBER is not needed and ignored.

N-Acetyl-D-glucosamine-6-sulfate ( 1' O and no 4' OH-group, Gaussian98, RESP)

AGS  INT    1
CORR OMIT DU   BEG
  0.000000
    1 DUMM   DU   M    0  -1  -2     0.0000    0.0000    0.0000  0.000
    2 DUMM   DU   M    1   0  -1     1.0000    0.0000    0.0000  0.000
    3 DUMM   DU   M    2   1   0     1.0000   90.0000    0.0000  0.000
    4 C1     AC   M    0   0   0     0.0000    0.0000    0.0000  0.124693
    5 C2     CT   M    0   0   0     0.0000    0.0000    0.0000  0.032432
    6 C3     CT   M    0   0   0     0.0000    0.0000    0.0000  0.107401
    7 C4     CT   M    0   0   0     0.0000    0.0000    0.0000  0.044439
    8 C5     CT   M    0   0   0     0.0000    0.0000    0.0000  0.098935
    9 C6     CT   M    0   0   0     0.0000    0.0000    0.0000  0.031706
   10 N      N    M    0   0   0     0.0000    0.0000    0.0000 -0.466264
   11 O1     OG   M    0   0   0     0.0000    0.0000    0.0000 -0.430845
   12 O3     OH   M    0   0   0     0.0000    0.0000    0.0000 -0.658535
   13 O5     OS   M    0   0   0     0.0000    0.0000    0.0000 -0.386715
   14 O6     OS   M    0   0   0     0.0000    0.0000    0.0000 -0.399527
   15 C2N    C    M    0   0   0     0.0000    0.0000    0.0000  0.749408
   16 O2N    O    M    0   0   0     0.0000    0.0000    0.0000 -0.632137
   17 CME    CT   M    0   0   0     0.0000    0.0000    0.0000 -0.448969
   18 HME    HC   M    0   0   0     0.0000    0.0000    0.0000  0.119617
   19 HME    HC   M    0   0   0     0.0000    0.0000    0.0000  0.119617
   20 HME    HC   M    0   0   0     0.0000    0.0000    0.0000  0.119617
   21 S      SO   M    0   0   0     0.0000    0.0000    0.0000  1.131685
   22 O1S    O2   M    0   0   0     0.0000    0.0000    0.0000 -0.603269
   23 O2S    O2   M    0   0   0     0.0000    0.0000    0.0000 -0.603269
   24 O3S    O2   M    0   0   0     0.0000    0.0000    0.0000 -0.603269
   25 H1     HC   M    0   0   0     0.0000    0.0000    0.0000  0.160956
   26 H2     HC   M    0   0   0     0.0000    0.0000    0.0000  0.127396
   27 H3     HC   M    0   0   0     0.0000    0.0000    0.0000  0.137259
   28 H4     HC   M    0   0   0     0.0000    0.0000    0.0000  0.139063
   29 H5     HC   M    0   0   0     0.0000    0.0000    0.0000  0.086498
   30 H6     HC   M    0   0   0     0.0000    0.0000    0.0000  0.085794
   31 H6     HC   M    0   0   0     0.0000    0.0000    0.0000  0.085794
   32 HN     H    M    0   0   0     0.0000    0.0000    0.0000  0.296182
   33 HO3    HO   M    0   0   0     0.0000    0.0000    0.0000  0.434306

LOOP
O5 C1

IMPROPER
 C2N  C2   N    HN
 CME  N    C2N  O2N

DONE

• Update the force field by starting YASARA with the command line option -upd:

  
  yasara -upd
  
  

YASARA will then recompile all force fields and tell you if something went wrong. If you do not get an error message, restart YASARA and try to initialize the simulation again. If YASARA still complains, consider the following points:

• Numbering of chemically equivalent hydrogens. The AMBER force fields do not follow the PDB convention that hydrogens bound to the same atom are numbered in the first column. YASARA corrects this problem for standard residues, but cannot guess the right answer for 'your' new residues. To avoid problems, do not additionally number hydrogens that are bound to the same atom. If you look at the example above, atoms 30 and 31 are bound to C6, and are both named H6, and neither H61/H62 nor 1H6/2H6.

• If you created a topology for an unusual amino acid by copying from a standard residue, you also have to delete the numbers of chemically equivalent hydrogens as just described above.

[1] Fast, efficient generation of high-quality atomic charges. AM1-BCC model: II. Parameterization and validation Jakalian A, Jack DB and Bayly CI (2002) J Comput Chem 23,1623-1641

[2] Development and Testing of a General Amber Force Field Wang J, Wolf RM, Caldwell JW, Kollman PA and Case DA (2004) submitted.

[3] MOPAC: A semiempirical molecular orbital program Stewart JJP (2000) J.Comp.Aided Mol.Des. 4,1-103

[4] Conductor-like screening model for real solvents: a new approach to the quantitative calculation of solvation phenomena Klamt A (1995) J.Phys.Chem. 99, 2224-2235