hi
this looks to me like the difference between r-6 avg and r-6 sum. Also
analysis is individually calculating the r-6 avg for hz2 -hg2* (5.219) and
hd1 -hg2* (7.465) and then taking the mean of those two numbers (6.342) and
comparing this to the target distance (5). This last bit of dividing the
restraint up into bits is something I don't think any of xplor/cns/aria do
so it possibly not very helpful.
for this kind of ambiguous restraint I would have thought you should either use
r-6 avg = {1/n x sum(r^-6)} ^ -1/6
(where ^ means 'to the power of')
which for your example gives something like 5.75 (so would give you a
violation of 0.75)
whereas
r-6 sum = {sum(r^-6)} ^ -1/6
which for your example gives something like 4.27 (as you said)
xplor/cns can use either r-6 sum or r-6 avg or a variety of other options
like pseudo atom etc (you have to choose in anneal.inp I think, and I don't
know what the default is).
aria uses r-6 sum always, because for highly ambiguous peaks with lots of
assignments (as in during an early aria iteration) where only 1 or 2 are
correct, the target distance will be skewed unfairly by all the incorrect
possibilities giving you a very small 1/n.
so if you used r-6 sum to calculate your structure then technically the
restraint is not violated.
tim perhaps you could introduce an option for this?
bye,
simon
|