hydrogens vs rescoring

Viewing 2 reply threads
  • Author
    Posts
    • #26899
      Christian Le Gouill
      Guest

      Hello,

      I used Gypsum-DL to create a library of 3D structures and screened them with Vina, Ledock and PLANTS. I would like to use Cyscore, Xscore, Ligand2, NNscore2… to rescore the VS results and do a consensus by compiling their Zscores.

      In Cyscore it is mentioned: "Please note: All Hydrogens of proteins and ligands are needed for correct calculation". However, I have in my library, files that have no hydrogen in them. I created them with Gypsum-DL, the SD file was cleaned using Datawarrior, and mol2 files extracted from the SD files using OpenBabel (obabel *.sdf -omol2 –split "Structure ID" ). I used the MGL tools script: prepare_ligand4.py (MGLTools) to convert to pdbqt files that would be compatible with Vina/NNScore2.

      I did not remove Hydrogen from the structures. However, when I used Fix2mol to prepare my ligands for xScore1.3, I get this message: "Warning: hydrogen atoms may not be correctly added. This molecule is skipped.".

      Is there any problem with the files I used to do VS or, I just need to add Hydrogen with Openbabel, post screening, before rescoring?

      I pasted the mol2 file content of one of these files.

      Best Regards,
      Christian

      REMARK 4 active torsions:
      REMARK status: (‘A’ for Active; ‘I’ for Inactive)
      REMARK I between atoms: C_2 and N_3
      REMARK 1 A between atoms: C_2 and C_4
      REMARK 2 A between atoms: C_13 and C_16
      REMARK 3 A between atoms: C_22 and C_24
      REMARK 4 A between atoms: C_24 and N_25
      ROOT
      ATOM 1 O UNL d 1 -1.951 -3.906 -1.330 0.00 0.00 -0.266 OA
      ATOM 2 C UNL d 1 -1.030 -3.118 -0.985 0.00 0.00 0.277 C
      ATOM 3 N UNL d 1 -1.350 -1.767 -0.613 0.00 0.00 -0.315 N
      ATOM 4 C UNL d 1 -2.566 -1.506 0.165 0.00 0.00 0.105 C
      ATOM 5 C UNL d 1 -0.747 -0.648 -1.349 0.00 0.00 0.105 C
      ATOM 6 C UNL d 1 -2.330 -0.339 1.129 0.00 0.00 0.032 C
      ATOM 7 C UNL d 1 -0.502 0.538 -0.410 0.00 0.00 0.032 C
      ATOM 8 C UNL d 1 -1.784 0.884 0.369 0.00 0.00 0.063 C
      ENDROOT
      BRANCH 2 9
      ATOM 9 C UNL d 1 0.357 -3.609 -0.946 0.00 0.00 0.173 A
      ATOM 10 N UNL d 1 0.744 -4.798 -1.473 0.00 0.00 -0.054 NA
      ATOM 11 C UNL d 1 2.058 -4.841 -1.179 0.00 0.00 0.735 A
      ATOM 12 N UNL d 1 2.896 -5.855 -1.505 0.00 0.00 -0.059 NA
      ATOM 13 N UNL d 1 2.494 -3.769 -0.504 0.00 0.00 -0.136 N
      ATOM 14 C UNL d 1 4.208 -5.770 -1.135 0.00 0.00 0.119 A
      ATOM 15 C UNL d 1 4.681 -4.649 -0.424 0.00 0.00 0.041 A
      ATOM 16 C UNL d 1 3.781 -3.624 -0.106 0.00 0.00 0.100 A
      ATOM 17 C UNL d 1 1.433 -2.971 -0.339 0.00 0.00 0.123 A
      ENDBRANCH 2 9
      BRANCH 8 18
      ATOM 18 C UNL d 1 -1.599 2.103 1.250 0.00 0.00 0.143 A
      ATOM 19 N UNL d 1 -2.164 3.287 0.938 0.00 0.00 -0.134 NA
      ATOM 20 N UNL d 1 -1.819 4.151 1.878 0.00 0.00 -0.132 NA
      ATOM 21 C UNL d 1 -1.043 3.488 2.755 0.00 0.00 0.177 A
      ATOM 22 N UNL d 1 -0.860 2.201 2.383 0.00 0.00 -0.299 N
      ATOM 23 C UNL d 1 -0.172 1.157 3.146 0.00 0.00 0.133 C
      BRANCH 21 24
      ATOM 24 C UNL d 1 -0.444 4.142 3.973 0.00 0.00 0.315 C
      BRANCH 24 25
      ATOM 25 N UNL d 1 0.961 4.553 3.742 0.00 0.00 0.149 N
      ATOM 26 C UNL d 1 1.038 5.702 2.820 0.00 0.00 0.286 C
      ATOM 27 C UNL d 1 1.601 4.900 5.022 0.00 0.00 0.286 C
      ENDBRANCH 24 25
      ENDBRANCH 21 24
      ENDBRANCH 8 18
      TORSDOF 4

    • #26905
      Christian Le Gouill
      Guest

      I pasted the pdbqt file and not mol2…

    • #27404
      Jacob Durrant
      Keymaster

      Hi Christian. Sounds like an interesting project! The PDBQT format includes only polar hydrogen atoms. If I’m not mistaken, the original AutoDock docking program only considers polar hydrogens, which is probably the motivation. My understanding is that AutoDock Vina doesn’t even use the polar hydrogen atoms explicitly, beyond using them to initially assign atom types to the heteroatoms. But I think the PDBQT format retains the polar hydrogens for historical reasons and for AutoDock4 compatibility.

      The problem with using Open Babel to reprotonate the docked ligands is that it will essentially undo the protonation states that Gypsum-DL assigns. Really what you need to do is only add back only the non-polar hydrogen atoms (e.g., ones bound to carbon atoms). I’m not aware of a program that can do that, unfortunately. It would probably take some Python coding, using Pybel or RDKit.

      Best of luck.

      • #27596
        Christian Le Gouill
        Guest

        Hi Dr Durrant,

        Thank you for your reply. At least my Vina results should be ok. I’ll try to rescore with prg that support pdbqt. However, the choices are a bit limited as most of them support mol2 files. When I try to convert Vina’s result to mol2 using openbabel,, I get this message for some of the poses:"Failed to kekulize aromatic bonds",.. so I’m stuck with pdbqt.

        My original plan was to screen with Vina/Qvina, PLANTS and Ledock, extract the the top 3 poses per docked structure, rescore the resulting 9 poses and compile a consensus. If I can’t reliably convert pdbqt files to mol2 files, I’ll rescore all poses I get from Vina, independently of the VS results from PLANTS and Ledock.

        Do you know other prg than openBabel for converting pdbqt directly to mol2?

        Thank you for your help and all the useful tools you developed.

        Best,
        Christian

      • #37837
        Jacob Durrant
        Keymaster

        Hi Christian. Sorry I can’t be more helpful. For what it’s worth, the RDKit library can load PDB files. Maybe it can also handle the PDBQT format. Take care.

Viewing 2 reply threads
  • The topic ‘hydrogens vs rescoring’ is closed to new replies.