witnotp &If you did not take a break, delete all molecules by delete mol * before you read in the file 4apr.mol. As mentioned in the previous section, saving the 4apr structure to a mol2 file lead to a reduction of the number of levels in the molecular data tree (from 5 to 4). Use the list command to confirm this:Wit!P> read mol 4apr.mol
Wit!P> list */*There is one node at level 1 (/4apr) with three direct descendants (E: "enzyme", I: "inhibitor", X: copy of I with alternative binding mode). The minimizer that will be used for the refinement of the binding modes can not handle multiple ligands in the same binding site (the two ligands would "see" each other through very strong repulsive vdW interactions), and it is therefore necessary to run separate refinements for the two binding modes:/4apr/E (mol)
/4apr/I (mol)
/4apr/X (mol)
Wit!P> buildThe energy refinements will be run on /mode1 (enzyme + inhibitor, experimentally observed binding mode) and /mode2 (enzyme + inhibitor, manually docked "upside down"). The original coordinates will be kept in /4apr for easy comparison of the structures before and after energy refinement.
Build> copy /4apr /mode1
Build> delete molecule /mode1/X
Build> copy /4apr /mode2
Build> delete molecule /mode2/I
The first structure that will be refined is /mode1. Since the other two structures make atom picking difficult (it is impossible to pick atoms that are overlapped in the display), it is a good idea to turn the display of molecules /4apr and /mode2 off:
Build> display molecule off 4apr mode2The energy refinement will be done by minimax:
Build> minimax molecule /mode1and use constraints on the atom positions of the enzyme:
minimax option: constraints new enzyme.fixi.e. all atom positions of the enzyme will be fixed, except those that are in a 10 Å "zone" around the inhibitor (if you are confused by the atom selection, consider re-doing the atomselection tutorial). The partial atomic charges of the molecule were determined using the MPEOE method. These charges work fine for in-vacuo calculations with the Tripos Associates Force Field (TAFF), if a distance dependent dielectric of 4.0 is used for the electrostatics:
add constraint> fixed /mode1/E -not /mode1/E -zone 10.0 /mode1/I -done -done
add constraint> done
minimax option: dielectric distance 4.0A significant speed-up of the computation of the force field energy of large systems may be achieved by neglecting long range interactions:
minimax option: cutoff 12 2 1i.e. computation of non-bonded interactions will be smoothly turned off over the switching distance of 2 A, interactions beyond 12 A will be ignored completely. To avoid the necessity to rebuild the non-bonded pair list in every iteration, pairs of atoms within cutoff+margin (12+1 A in the example) will be included in the non-bonded list, and the non-bonded list will only be updated if there are atoms which have moved more than 0.5 Å.
On multi-processor machines, another way to speed up the calculations would be to use several processors in parallel for the energy calculations, i.e. to use 6 processors:
minimax option: processors 6Asking for more processors than available is not a good idea: the program will try to dynamically adjust the number of parallel threads, but this is an expensive process that can take a quite some time (couple of minutes) during which the performance will be poor.
This completes the set-up of the energy refinement of the first binding mode. To start the actual calculation, type
minimax option: run mode1mode1 is a job name, and will be used as the root name of two files that will be generated in the current working directory (mode1.mol, with the molecule in MOL2 format, and mode1.dat, with a printable report of the progress of the energy refinement). The calculation will be done in the foreground, with updates of the displayed atomic positions every 10th iteration. The refinement of the first binding mode should run to completion (default rms gradient of 0.0003 mdyn) in about 650 iterations. The impatient may abort the minimization by clicking the abort menu button. Once completed the minimax command will return to the Build menu.
Minimax options are "sticky", i.e. once set, they remain set, until they are modified, or reset to their default values (by minimax reset). Therefore, to run the refinement of the second binding mode it is sufficient to
Build> display molecule toggle mode1 mode2The constraints that were used to fix most of the enzyme atom positions in the first run of minimax can be used for the second run, since mode1 and mode2 have the same atom numbering. If the starting conformation (from manual docking in the previous section) is very bad, the refinement may terminate before reaching convergence. In this case it may help to use the bad option of minimax:
Build> minimax molecule mode2
minimax option: run mode2
Build> minimax mode2If the calculation still does not run to normal completion, the starting binding mode needs to be improved manually.
minimax option: bad yes
minimax option: run mode2
The tutorial ends here, rather abruptly. In WWW parlance, this page is "under construction". For those familiar with Basel: the Nordtangente is "under construction" since the early seventies of the previous century, and for more than 25 years the Nationalstrasse A2 ended in mid-air. If all goes well, the last piece of Nordtangente will be finished in 2006. So, again, this page is under construction. Please be patient.