Getting Started

After the Installation and familiarizing with the most important Input Files and Output Files, the best way to get started with SEED is to try to replicate some Test Cases and start understanding the main parameters with the help of the Parameter File Generator.

Installation

SEED code is hosted on GitLab.

In order to build the latest development version of SEED do the following (you may have to modify the Makefile.local in the src folder):

git clone https://gitlab.com/CaflischLab/SEED
cd SEED/src
make seed

SEED makes use of the BOOST C++ Libraries. The necessary header files are distributed along with SEED. The binary is compiled into the SEED/bin folder.

After installation run SEED with the following command:

seed_4 seed.inp >& log

For compiling and running the parallel version of SEED, refer to Running SEED on a cluster.

Input Files

The following is a list of the input files required for a SEED run:

  • The file seed.inp contains the most frequently modified input values (path and name of structural input files, list of residues forming the binding pocket, switch between polar and apolar docking, …). For detailed information see Input Parameters.

  • The parameter file seed.par (filename can be specified in i1) contains less frequently modified input/output options, parameters for docking, energy and clustering. Modification of some parameters in this file is recommended only to advanced users who wish to fine tune the SEED energy model.

    Both seed.inp and seed.par allows comment lines (starting with a #) and the files must terminate with the word end. All the other lines hold information read by the program. It is important that this information is provided in the correct order for parsing by the program. In this documentation and in the pdf user manual, parameters referring to the input and parameter files are indicated by i and p respectively. For a detailed description refer to Parameters. For a full working example of the input and parameter file try out the Parameter File Generator and the test cases in the Tutorials section.

  • A standard SYBYL mol2 format file containing all the fragments to dock. This file is simply the concatenation of all the fragments, expressed in mol2 format. Note that different conformations of the same fragment are treated as different fragments. Partial charges are written in the 9th column in the @<TRIPOS>ATOM record.

    In order to retrieve the correct van der Waals parameters from seed.par file (block p29), the CHARMM atom types should be specified in the mol2 file. This is done using the alternative atom type specified by the record @<TRIPOS>ALT_TYPE, which takes the following form:

    @<TRIPOS>ALT_TYPE
CGenFF_4.0_ALT_TYPE_SET
CGenFF_4.0 1 CG331 2 CG301 3 CG331 4 CG324 ...

    Where CGenFF_4.0_ALT_TYPE_SET sets a user-defined name (for example CGenFF_4.0) for the alternative atom type set. This name is repeated on the next line, followed by the list of “atom number-atom type” pairs for each atom in the molecule. This list should span a single line, but can be broken by using \\. It is recommended to keep the SYBYL atom types on the 6th column of the record @<TRIPOS>ATOM as they are recognized by most cheminformatics and visualization software. The first line of the SYBYL record @<TRIPOS>MOLECULE specifies the fragment name. It is convenient (but not necessary) to have unique names for each fragment. In case fragments with duplicate names are found in the input, they will be renamed in all the output files appending to their name the dollar sign $ and an incremental index. As the fragment mol2 input file is read sequentially, the number of fragments in it does not have to be specified a priori.

  • A standard SYBYL mol2 file for the receptor with partial charges on the 9th column in the @<TRIPOS>ATOM record (as for the fragments) and CHARMM atom types specified by the @<TRIPOS>ALT_TYPE record (refer to the fragment file description for details).

Output Files

The main SEED output file, whose filename is specified in p6 (by default seed.out), contains detailed information about the energy values (with both fast and accurate model) and results of clustering. The first term of p28 is the maximal number of lines that can be written in the main output file for each docking step of each fragment type. The second term of p28 gives control on which information may be discarded in the output file (print level).

A directory outputs in which all the output files are written is automatically created by the program. Note that if a directory named outputs is already present, it will be overwritten by the SEED run.

<FragmentMol2FileName>_clus.mol2 contains the fragment top poses per cluster ranked by accurate energy after the postprocessing step. This file is the concatenation of a mol2 file for each saved pose. The maximum number of poses to be saved per cluster can be set in p5 (first value). The comment line of the SYBYL mol2 record @<TRIPOS>MOLECULE (6th line after the record identifier) contains some useful information about the pose, i.e. increasing pose index, cluster number, total energy and fragment number (Fr_nu). The latter represents the program internal numbering of the pose and it is not interesting per se, but it can be used to match the pose to docking information written in seed.out.

seed_clus.dat is a summary table containing the separate energy terms for each fragment position saved to <FragmentMol2FileName>_clus.mol2. This information can be also retrieved from the main output file. Columns are organized as follows:

  • Name: Fragment name.
  • Pose: Incremental pose number. This index restarts at 1 for each new fragment.
  • Cluster: Cluster number.
  • Fr_nu: Fragment number. This is SEED internal pose number.
  • Tot: Total binding energy.
  • ElinW: Electrostatic interaction in water.
  • rec_des: Desolvation of the receptor upon complex formation.
  • frg_des: Desolvation of the fragment upon complex formation.
  • vdW: Van der Waals interaction energy.
  • DElec: Electrostatic difference upon fragment binding. It is given by ElinW-DG_hydr. It roughly represents how good the fragment feels in the protein compared to how good it feels in water.
  • DG_hydr: Free energy of hydration of the fragment.
  • Tot_eff: Tot/HAC.
  • vdW_eff: vdW/HAC.
  • Elec_eff: ElinW/HAC.
  • HAC: Heavy atom count. It is the total number of non-hydrogen atoms in the fragment.
  • MW: Molecular weight of the fragment.

<FragmentMol2FileName>_best.mol2 contains the best fragment positions, according to the total binding energy, irrespective of the cluster they belong to (maximum number of saved poses set by p5, second value). The difference with respect to <FragmentMol2FileName>_clus.mol2 is that the user can set the total number of poses to be saved instead of the number of cluster members.

seed_best.dat is the same as seed_clus.dat but matching <FragmentMol2FileName>_best.mol2.

The writing of the above *_clus.mol2 and *_best.mol2 files is activated or deacti- vated by p3 (first and second value respectively). The writing of the seed_clus.dat and seed_best.dat summary table is activated or deactivated by p4 (first and second value respectively). Note that the maximum number of poses and poses per cluster to be saved (p5) are upper bounds as the number of generated poses may be smaller than the number of poses requested in output. The four parameters for writing the output files (p3 and p4) can be switched on/off independently.

Note that the number of cluster members to be saved (first value of p5) implicitly determines the maximum number of poses for which to evaluate the accurate binding energy. Thus in general it is advisable to set this number to a value higher than one, in order to be sure to consider a meaningful number of poses, and to suppress the corresponding mol2 file output (first value of p3 set to n) as it may quickly become big.

Other output files

Besides the docking output files containing structural information and energy values, SEED generates some additional output files.

The grids for the evaluation of fast van der Waals energy, fast screened interaction energy and receptor desolvation can be saved on disk and reused for a subsequent run (see p7, p8, p9). The grid files are saved by default in the `scratch subfolder.

When a new project is started, it can be very useful to first generate and visualize the vectors used for ligand placement, before performing any docking (see Vectors for docking for details). Vectors are saved in the following mol2 files and can be opened in a molecular viewer:

  • polar_rec.mol2 contains vectors distributed uniformly on a spherical region around each ideal H-bond direction. The deviation from ideal hydrogen bond geometry and the number of additional vectors to distribute uniformly on the spherical region are set in p12.
  • polar_rec_reduc_angle.mol2 contains vectors of polar_rec.mol2 which are selected according to an angle criterion (i4, p14). Vectors pointing outside of the binding site are discarded. The file polar_rec_reduc_angle.mol2 exists only if the angle criterion has been activated by the user (i4).
  • polar_rec_reduc.mol2 contains vectors of polar_rec.mol2 (or of polar_rec_reduc_angle.mol2 if the angle criterion has been activated (i4)) which are selected according to favorable van der Waals interaction between all the receptor atoms and a spherical probe on the vector extremity. The aim is to discard receptor vectors that point into region of space occupied by other atoms of the receptor and select preferentially vectors in the concave regions of the receptor. The van der Waals radius of the probe is specified in p15. The number of selected vectors is controlled with p2.
  • apolar_rec.mol2 contains points distributed uniformly on the solvent-accessible surface of the receptor. The density of surface points is set in p22.
  • apolar_rec_reduc_angle.mol2 contains vectors of apolar_rec.mol2 which are selected according to an angle criterion (i4, p14). Vectors pointing outside of the binding site are discarded. The file apolar_rec_reduc_angle.mol2 exists only if the angle criterion has been activated by the user (i4).
  • apolar_rec_reduc.mol2 contains points of apolar_rec.mol2. (or of apolar_rec_reduc_angle.mol2 if the angle criterion has been activated (i4)) which are selected according to their hydrophobicity. For this purpose a low dielectric sphere is placed on each of these points. The hydrophobicity is defined as the weighted sum of the receptor desolvation energy due to the presence of the probe and the probe/receptor van der Waals interaction. The weighting factors and the probe radius are set in p22. The number of selected apolar points is controlled with p2.

Of the six files listed above one should visualize polar_rec_reduc.mol2 and apolar_rec_reduc.mol2. It is useful to modify the appropriate parameters if the vector distributions do not meet the user’s expectation, since fragments are docked using the vectors present in these files. As soon as the you are happy with the generated vectors, you can just read the maps (first value of p7, p8, p9 set to r) instead of generating and writing them again (first value set to w).

The file sas_apolar.pdb contains points defining the solvent accessible surface of the binding site, which can be visualized with a molecular viewer.

Troubleshooting

If after starting a SEED run the program exits unexpectedly, the keyword WARNING should be looked for in the main output file (seed.out, p6) to find hints on possible problems (wrong path for filenames, unknown value for some parameters, …).

The docking workflow implemented in SEED involves many filtering steps, hence, if the main output file does not contain any fragment position for a given fragment type, it can be due to several reasons: the center of the spherical cutoff (i6) might be misplaced (outside the binding site), the checking of steric clashes (p10 and p11) too strict, the van der Waals energy cutoff (p19) for apolar fragments too severe, the total energy cutoff (third value of i7), or the energy cutoff for the second clustering (fourth value of i7) too stringent. To find out what the reason could be, the following part of the main output file should be investigated:

Total number of generated fragments of type 1 (BENZ) : 118800
Fragments that passed the sphere checking : 102894
Fragments that passed the bump checking : 49007
Fragments that passed the vdW energy cutoff : 22100
Fragments that passed the total energy cutoff : 17794

Parallel SEED

SEED has an MPI-parallel version. If you want to run a SEED screening campaign on a cluster, refer to Running SEED on a cluster.

MC minimization

Stochastic minimization with a rigid-body Monte Carlo Simulated Annealing scheme can be enabled to run on the top generated poses; these are the poses for which the accurate binding energy is evaluated and their maximum number (per cluster) of can be tuned with the first value of p5.

MC minimization can be used in both Docking running mode and Energy evaluation mode. The latter can be useful when rescoring poses not generated by SEED. In this case, if you had to run rescoring without minimizing the poses, you might get very unfavourable energy values due to the ruggedness of the van der Waals energy landscape.

If you want to know more details, refer to Monte Carlo Simulated Annealing.