The Input File

This section outlines the input file ofQMPBplacing emphasis on the general structure, closely reflecting the object structure discussed in the next section, and leaving a full explanation for each possible input option to appendix A.6. The aim of defining the input file format was to allow access to the full functionality of QMPB. Since it was thought to be written by pre-processors, it was not important to be very compact. Instead it should be well readable by the user to check and modify input options. The input is structured into a block with general options and blocks for each instance of each site. The options are mostly given askey=value pairs, where the value can sometimes be a space separated list of values. The order of the parameters in a block and the order of blocks is arbitrary. I refer to line numbers of the given examples in the following, however the line numbers are generally not important for the program. All energies are given in units of ^kcal_mol.

The General Block

The general block may contain options similar to the following example:

1 meadpath = /home/essigke/bin 2 T = 300

3 I = 0.1

4 b a c k f i l e = background . pqr 5 MGMcenter = ON CENT OF INTR 6 MGMpoints = 131

7 MGMspace = 0.2

8 OGMcenter = ON GEOM CENT ON CENT OF INTR 9 OGMpoints = 131 131

10 OGMspace = 1 0.2 11 epsin1 = 1 12 epsin2 = 4

13 Ligand Labels = proton e l e c t r o n

Line 1 in the example defines the path to the helper programs (section 4.4), then some param-eters for the Poisson-Boltzmann solver are given (absolute temperatureT and ionic strength I). Line 4 gives the name of the background pqr-file. It contains the coordinates, charges and

radii of all atoms of the molecule not belonging to any site. The next six lines specify parame-ters for the grid, on which the PBE is solved numerically. The options starting withMGMdefine parameters for the grid of model compounds (or Modelsite), the other define parameters for the grid of instances of sites (the terminology is adopted fromMEAD). It is shown, that either a single or multiple values can be given. As many focussing steps are done when solving the PBE as there are values. In a similar way, grids can also be specified for instances (in the blocks below, however only oncepersite) overwriting the default values given here. The option to use different grid definitions allows using a rather small grid (which is fast to solve) for most sites, but larger grids for sites, which require it.

The centering type allows three options (as in Multiflex). ON ORIGIN defines the center of the grid to be at the origin (0,0,0). ON GEOM CENT defines the center of the grid to be at the geometrical center of the instance pqr-file. Both options are translated into coordinates by theMEADlibrary. The third optionON CENT OF INTR(on center of interest) defines the center of the grid to be at the geometrical center of all instances of a particular site. This center is therefore identical for all instances of the site as it is required to cancel grid artefacts. It is at an optimal position to allow small grids with high resolution. Since calculation of this center requires knowledge of the coordinates of all atoms in all instances, it is done by QMPB. The keyword is replaced by the calculated coordinates in the grid files written by QMPB for all instances.

The keys epsin1, epsin2and (optionally) epsext define the dielectric constant of the three dielectric regions currently available inQMPB. In the current version ofQMPBthey are manda-tory, but in future versions they probably will be removed in favor of the eps key in the in-stance blocks. A list of names for the ligand types are assigned to the keyLigand Labelsin line 13. The names have to be given in the same order as the values for the keysNandGfree discussed below.

An Instance of a QMsite

The next example shows the definition of an instance of a site, for which absolute binding energies should be calculated (section 3.3). The instance can be parameterized by quantum mechanical (QM) calculations, explaining the nameQMsite.

1 QMsite site C PHE 39 A instance 0 ox 2 f i l e =site C PHE 39 Ainstance 0 ox . pqr

3 sid=21

4 i i d =0

5 eps= 1

6 Hqm=−14998.1 7 Gvib=98.693

8 N= 0 0

9 Gfree= 0 −102.1558

10 QM corr C PHE 39 A . pqr int C PHE 39 A . pqr 11 QM corr O PHE 39 A . pqr int O PHE 39 A . pqr 12 . . .

Line 1 specifies the type of site, a unique label for the site and a unique label for the instance.

These labels should be intuitive for the user, facilitating a quick understanding of the input and output files, where these labels are reused. Line 2 sets the name of the pqr-file, which is defining the coordinates, charges and radii of atoms belonging to the instance of the site.

Line 5 defines the dielectric constant of the region of the site. The optional keys sid and iid (line 3 and 4) associate the site and instance with unique numbers. They are used for sorting the output, which is usefull for post-processing scripts, but also comparison of different calculations. If the keys sidandiidare omitted, they are automatically generated byQMPB. Line 6 and 7 define the quantum chemical energy of the instance (Hqm, total bonding energy or energy of formation,H_vac,i(j_k)in eq. 3.32) and the vibrational energy (Gvib, which is Gvib,i(jk)in eq. 3.32), respectively. Line 8 and 9 give the number of bound ligands of each type (N, which isn_λ,i(j_k)in eq. 3.7 and eq. 3.34) and the standard chemical potentials for all ligand types (Gfree, which is µ^◦_λ in eq. 3.34). Line 10 and the following lines are for calculating the correction energy (G_corr,i(j_k) in eq. 3.32). Section 3.5.2 explains, why this correction is necessary. The first pqr-file in eachQM corrline specifies an atom in theQMsite(named Q1 or Q2 in Fig. 3.7) and the second pqr-file specifies the atoms outside of this site, with which it artificially interacts (M1 and M2 in Fig. 3.7). Alternatively to theQM corrlines, a line with the keyGcorrand the correction energy as value, can be given. Calculating the correction energy is very expensive and usually redundant in repetitive calculations.

An Instance of a MMsite - Non-Ligand Binding Reference Rotamer Form

Section 3.4 describes the calculation of energies of instances of sites relative to a reference.

Four types of such instances were distinguished. The energy of the non-ligand binding ref-erence rotamer form is defined to be zero. Nevertheless, the electrostatic energy in the het-erogeneous environment of the molecule compared to a homogeneous dielectric is required by non-reference rotamer forms. Therefore the energy is calculated.

1 MMsite s i t e HT1 ALA 1 A instance 0 C 2 f i l e = s i t e HT1 ALA 1 Ainstance 0 C. pqr

3 sid=0

4 i i d =0 5 r e f = s e l f

6 eps= 4

7 N= 0 0

8 Gmm=0.3299

As for the QMsite, line 1 specifies the type of site, a unique label for the site and a unique label for the instance. The nameMMsitewas chosen in contrast toQMsite. Often the energies of rotamers of this type of site (given in line 8) are obtained from a molecular mechanics (MM) force field. Line 2-4 and 6-7 are analogous to theQMsite. The keyrefin line 5 has the value self, indicating that the reference instance of this instance is the instance itself, neither anotherMMsitenor aModelsite.

An Instance of a MMsite - Non-Ligand Binding Non-Reference Rotamer Form

The next example shows the input for an instance, which takes the previous example as reference rotamer:

1 MMsite s i t e HT1 ALA 1 A instance 1 A 2 f i l e = s i t e HT1 ALA 1 Ainstance 1 A . pqr

3 sid=0

4 i i d =1

5 r e f =instance 0 C 6 Gmm=0.3355

The key ref in line 5 now has the value instance 0 C, which is the instance label of the previous example instance. By that the program knows, that it should use instance 0 C as reference instance for instance 1 A. The rotamer energy in line 6 is different from the reference instance. Note, that certain keys can only be given to the reference instance, not to non-reference instances.

An Instance of a MMsite - Ligand Binding Reference Rotamer Form

In the previous examples, instances of typeMMsiteonly differed in their coordinates (rotamer form), not in the number of ligands bound (charge form). In this case, the ligand binding is described by a model compound in solution, for which ligand binding energies are known (e.g.,from experiment).

1 MMsite s i t e CE LYS 4 A instance 0 p ROT−5 2 f i l e = s i t e CE LYS 4 Ainstance 0 p ROT−5.pqr

3 sid=1

4 i i d =0

5 r e f =model instance 0 p ROT−5 6 Gmm=0.0316

The key refin line 5 now has the value model instance 0 p ROT -5, which is the instance label of the instance of aModelsite. If the site only exists in a single rotamer form, the option nohomoshould be given for theMMsite. This option avoids the calculation of the instance in a homogeneous dielectric, which is only needed for rotamers.

An Instance of a Modelsite

The next example shows how to specify a model compound:

1 Modelsite s i t e CE LYS 4 A model instance 0 p ROT−5 2 f i l e =model s i t e CE LYS 4 Ainstance 0 p ROT−5.pqr

3 sid=1

4 i i d =0

5 r e f =instance 0 p ROT−5 6 Gmodel=−14.2665506152

7 eps= 4

8 N= 1 0

The input is very similar to a non-ligand binding reference rotamer form of a MMsite. The type of site is nowModelsite. A rotamer energyGmm is not specified, because it is specified for the associatedMMsitegiven by the keyref. Instead, the energy of the model compound in solution is specified with the keyGmodel. It is important, that allModelsiteinstances of a site (due to different charge forms) are in the same rotamer form. The file specifying theModelsite has to contain all atoms (with identical coordinates, charges and radii) as the associated MMsite, but usually contains additional atoms, which were present when determining the model energy in solution. The dielectric boundaries are calculated from all atoms in this file. The additional atoms contribute with their charges as background charge set to the electrostatic energy.

An Instance of a MMsite - Ligand Binding Non-Reference Rotamer Form

Finally, there might be rotamer form, which takes a ligand binding rotamer form as reference:

1 MMsite s i t e CE LYS 4 A instance 1 p ROT 24 2 f i l e = s i t e CE LYS 4 Ainstance 1 p ROT 24. pqr

3 sid=1

4 i i d =1

5 r e f =instance 0 p ROT−5 6 Gmm=0.6172

The input is analogous to a non-ligand binding non-reference rotamer form, but the keyref now points to instance instance 0 p ROT -5. The distinction, necessary for the different calculation schemes can only be made by analyzing the refvalue of the reference instance.

If it is self it is a non-ligand binding non-reference rotamer form, instead if a Modelsite instance is specified, it is a ligand binding non-reference rotamer form.