Analysing the results

From DISI
Revision as of 14:53, 3 November 2009 by Kolb (Talk)

Jump to: navigation, search

Some analyses that can be performed

Combining the results of all subdirectories

  • in the subdirectory that contains all the individual directories for each chunk of the library, run combine10.csh FF `pwd`, where FF is the common part of the subdirectories you want to combine. This should be done on sgehead, since the script needs to access the ZINC database. The output consists of an energy list (FF.2energy) and corresponding FF.new.eel1 file as well as a list of the charges and heavy atom counts (FF.new.chg).
  • if one wants to create an .eel1 file for a certain subset of the molecules, first create the list of molecule names plus their energies (on one line) and then feed it to getxpdb.pl name_energy.list < FF.test.eel1 > subset_name.eel1.

Getting individual atom contributions with scoreopt_so

Converting a .mol2 file into an .eel1 file

  • run amsol as described here to calculate atomic solvation energies.
  • run file2file.py -s path/to/amsol.solv path/to/amsol.nmol2 ligand.eel1.

Individual contributions to the coulombic energy

  • start scoreopt_so and choose option '2' in the first menu.
  • enter the name of the DelPhi potential file, presumably grids/rec+sph.phi.
  • enter the name of the ligand file, i.e., ligand.eel1 .
  • enter the name of the output file, e.g. ligand.elec .
  • in every ATOM line, columns 9, 10 and 11 are the partial charge, the electrostatic field and the energy in kT (i.e., 9 × 10) of the atom, respectively.
  • the DelPhi electrostatic score is the sum over the entries in column 11 times 0.5924 (conversion from kT to kcal/mol) and can be compared to the elect column in OUTDOCK.

Individual contributions to the van der Waals energy

  • start scoreopt_so and choose option '3' in the first menu.
  • enter the prefix name of grids for ff scoring as a full path, i.e., grids/chem .
  • enter the name of the van der Waals parameter file, presumably grids/vdw.parms.amb.mindock .
  • answer the question about interpolation with 'yes'.
  • enter a sufficiently large number as maximal van der Waals energy, e.g. 10000.
  • enter the name of the ligand file, i.e., ligand.eel1 .
  • enter the name of the output file, e.g. ligand.vdw .
  • be adequately scared.
  • the van der Waals interaction energy is calculated as <math>{vdW}_{(r)}=\frac{A}{r^{12}}-\frac{B}{r^6}=a-b</math>. In every ATOM line, columns 9, 10 and 11 are <math>a</math>, <math>b</math> and <math>a-b</math>, respectively.
  • DO NOT use the interaction energy, as we only use the vdw component now. Instead, use the vdwsum to compare with the vdW column in OUTDOCK.

Individual contributions to the desolvation

  • start scoreopt_so and choose option '4' in the first menu.
  • enter the name of the grid for partial desolvation, presumably grids/solvmap .
  • enter the name of the ligand file, i.e., ligand.eel1 .
  • enter the name of the output file, e.g. ligand.solv .
  • in every ATOM line, columns 9, 10, and 11 are the total atomic solvation energy (polar + apolar), percentage desolvation, and atomic desolvation energy (i.e. - 9 × 10) of the atom, respectively.
  • the total desolvation is the sum over the entries in column 11 and can be compared to the sum of the polsol and apolsol columns in OUTDOCK.

Other small useful things

Obtaining the net charge of a docked molecule

  • take the output .eel1 file and run molcharge_pdb.pl < output.eel1. This will output the sequential number of the molecule, the ZINC identifier, the total charge and the number of atoms for every molecule in the file. This script is called by combine10.csh and the output is called FF.new.chg (cf. section 5.1).