Dave's decision is good that we keep it as is for this release.
Adrian (and others) is absolutely right. Indeed 10-20 ns is essentially too
short for anything to happen if the goal is to study large scale complex
dynamics. But, it happens that some users may not be interested in the large
scale complex motion. Rather, they may want to know local motions and they
do not take wait-and-see approach.
Here are two hypothetical studies to illustrate what I mean.
Suppose you have a large system that is essentially stable. But there are a
bunch of small mobile parts (e.g., side chains and surface loops) and these
small parts are thought to play key functional roles. It also happens that
X-ray data is not accurate enough and these small moving parts are
disordered in the high-resolution structures (because they are mobile). Your
fiends ask you "can you try to figure out how these little guys get involved
in functions"?
Suppose you have a large oligomeric system, your goal is to investigate the
different modes to form the complex and you anticipate (from other evidence)
that the proteins to do not change much upon oligomerization. You decide to
try a bunch of simulations to investigate the possible binding modes
(obtained by other means), all by 10-20ns simulations. The only problem is
that the system is a bit large, as typical case for complexes.
I feel that it would be nice to enable these types of studies if possible
and if the cost is small.
yong
Received on Sun Dec 09 2007 - 06:07:15 PST
