Documentation for the HOLE: 8.0 predicting conductance properties of ion channels

Navigation:

Back to previous section 7.0 Ancillary programs

8.0 Using HOLE to predict conductance properties of ion channels

This is a major new feature in release 2. The science behind the method is described in:

Smart, O. S., Breed, J., Smith, G. R. and Sansom, M. S. P. (1997): A novel method for structure-based prediction of ion channel conductance properties. Biophys. J. (March issue).; or A brief description can be found in the J. Mol. Graphics paper included in this documentation as section 2.2 of this documentation.

This section will concentrate on the practicalities of how to perform conductance prediction calculations on your system.

8.1 Absolute conductance predictions in KCl

The following input file is taken for one of the runs which was used to parameterize the correction functions used to make empirical corrections to the macroscopic expected conductance calculated by HOLE. For meaningful corrections the conditions used should be followed reasonably carefully.

! calculate conductivity for porin
coord pdb1omf.pdb               ! Co-ordinates
radius ~/hole2/rad/simple.rad
cvect  0 0 1                    ! channel runs approx up Z axis
cpoint -1.17 49.45 33.72        ! point in pore
sample 0.25                     ! distance between planes
capsule
ignore hoh
shorto 1
mcdisp 0.4
mcstep 10000
sphpdb test.sph

Notes:

The prediction routines are turned on as part of the capsule option which measures the anisotropy of a channel.

You MUST use the non-united atom AMBER van der Waals radii set included as simple.rad.

You must exclude from consideration all waters, ions etc. within your channel (done with an ignore card here).

The mcstep 10000 means that ten times as many steps as normal are done in the Monte Carlo Optimization. Although this results in long runs on all but the fastest work station it is necessary to achieve consistent results (in terms of area) between runs. If you are doing multiple files (e.g. averaging the result along a dynamics run) then you can reduce the number to 3000 but I wouldn't go any lower.

The value of sample should not be critical but I would use 0.25 for consistency.

The value of mcdisp should be set to about 1/10th of the minimum effective radius of the channel you are looking at.

The SPHPDB card produces a file which you can use with the sphqpt program to visually check the performance of HOLE at the end of the run. It is a good idea to do this to make sure the program is not doing anything silly.

Results of the conductance prediction are reported at the end of normal HOLE output (written to standard output but normally redirected to a file). These are reasonably self-explanatory. This output is taken from run on porin 1omf using input file above. The run took at total of 1hour 10mins cpu on an IRIX indigo2 R4400:

 The geometric factor F= sum(ds/area) along channel is    0.344 angstroms**-1
 This yields a macroscopic predicted molar conductance of:
   (1/rho)*(100/F)= (290/rho) pS,
   where rho is the conductivity of 1M permeant ion in ohm m.
 For 1M KCl rho= 1/12 (ohm m), So Gmacro= 3500 pS/M.
 Empirically derived correction factors:
    first generation (2 system) on MINRAD=      4.97989
    2nd generation (8 system) constant=         5.59400
    2nd generation on length of channel=        5.80046
    2nd generation on avg elect potential=      6.14517
      0.00000

    first generation (2 system) on MINRAD= 700 pS/M.
    2nd generation (8 system) constant=    620 pS/M.
    2nd generation on length of channel=   600 pS/M.
    2nd generation on avg elect potential= 570 pS/M.
 
  As a single estimate take average of last 3 so Gpred= 600 pS/M.
 N.B., The predicted conductances given above are only
       a crude estimate and should not be over interpreted!

The experimental conductance of 1omf in 1M KCl is 700 pS (the calculated result is not that surprising considering this was included in the parameterization of the model!).

If you want to make a prediction for an ionic concentration other than 1M then simply multiply the predicted conductances by the molar concentration wanted (i.e. for 0.5M KCl multiply all the numbers by 0.5).

Please let me know how you get on.

8.2 Predicting the effect of adding non-electrolytes to conductance experiments

The science behind this is explained in the Smart et al. (1997) paper. The calculation routines are started by a PEGRAT card:

pegrat 0.54 1.18 ! produce PEG graph * 20% PEG

The numbers following the card specify the value of theta_NON and theta_PEG as defined in Smart et al. (1997). The capsule option cannot be used with this. The calculations are in a very early state of development. Please get in touch with me if you are interested in using these. YOU MUST CHECK WITH ME BEFORE YOU PUBLISH ANY RESULTS USING THIS ROUTINE.

Navigation:

Back to contents

Back to previous section 7.0 Ancillary programs

On to next section 9.0 Example run