On Aug 27, 2007, at 4:36 AM, Jakob Toudahl Nielsen wrote:
> Dear Xplor-nih user, (replying my own email)
>
> I switched from xplor-nih version 2.16 to version 2.18. Now my
> convergence problems are solved!
Glad to hear it.
> I tried the PASD calculation with the same input parameters again
> starting from pass3. I notice a few differences between the pass3
> proc in v2.16 and the new v2.18. I guess the reason could be that
> in the new version all VDW intactions are turned off during the
> high temperature conformational search phase whereas CA-CA VDWs are
> present in v2.16.
> Are there any other main differences?
The reason I turned off the vdWs during high temperature is that the
"pass3" protocol (which might be renamed soon, since I've dropped
down to two
passes of structure calculations in the current release) now can take
advantage of inverse NOEs, which seek to keep separate proton pairs
which don't
give rise to an NOE peak. This is a really valuable source of
information which is generally under-used.
> What is the meaning of "inactive peaks"? I thought that all peaks
> were active during pass3 and the precisely one peakAssignment per
> peak were active at the time. Is someting different in the new
> version?
Nothing about that has changed in the new version. It's always been
possible for a particular peak to be entirely inactive during pass3.
In pass3, activation of peakAssignments happens in two steps:
1. A single peakAssignment is chosen among all the
peakAssignments of each peak. All non-chosen peakAssignments are
inactivated.
2. Then the chosen peakAssignment is subjected to the same
activation/inactivation process used during pass2.
This allows us to deal with entirely bad peaks in a graceful manner.
Hope this helps. Many thanks to Charles, by the way, for watching
the fort while I was on vacation.
Maine is much nicer than Washington in August!
--JK
