Yes, I've seen this behavior. I attribute it to lack of the 0,0,0 and ultra-low
resolution reflections. Remember that the average density in a map (with
difference
coefficients or whatever) is zero. That means that if the model lacks a few
atoms,
but has no atoms that don't belong, so that an absolute difference map would go
from zero to some positive value, The calculated map would go from some negative
value to a lower positive value. Thus a point where there is no atom and should
be no atom , with a value of 0-0 = 0, may be the lowest point in the map and
come
out negative in the "AC-coupled" map.
That still wouldn't account for *significant* negative peaks, since the positive
density peaks are few and the near-zero points are many, the baseline will not
drop much to float the peaks. However if low-resolution data is missing, the
density profile is being put through a hi-frequency filter, so the whole
baseline
is not shifted down equally, but rather the baseline in the region of the peaks
sinks under the weight of the peaks. So, out of all the points in the absolute
map that were near zero, a few points in the vicinity of the highest positive
peaks may end up lower than all the rest and appear significant.
Also, if you are judging "significant" by the old 3-sigma rule, remember that as you
improve your model the difference map tends to zero everywhere, and the absolute
value associated with 3 sigma gets smaller and smaller. Much better to look at
difference maps in terms of absolute density (e-/A^3; the default now in O), or
compare the sigma value of the negative peaks with that of the positive peak
produced by omitting a well-ordered water. Then they may not seem so
significant!
This explanation is not completely satisfactory, so if anyone has a better one
I'm eager to hear it.
Ed
Waight, Andrew wrote:
Hello everyone I have a question for the experts.
I am in the final stages of refining my model placing waters and whatnot. However when I refine my model against the pure scalepack output, I see some rather signifigant negative difference density peaks (3sigma) in a marginally important region of the protein. The strange thing is that there is no model built into this region. How can there be negative difference density when there is no model built there? It is obviously not a weighting problem because the structure factors are directly from the raw reflection list. Has anyone else ecperienced this phenomenon and what if any actions would you suggest. I have attached also a screenshot. Thanks for your advice everyone I have learned so much from reading this BB everyday.
Drew
------------------------------------------------------------
This email message, including any attachments, is for the sole use of the
intended recipient(s) and may contain information that is proprietary,
confidential, and exempt from disclosure under applicable law. Any unauthorized
review, use, disclosure, or distribution is prohibited. If you have received
this email in error please notify the sender by return email and delete the
original message. Please note, the recipient should check this email and any
attachments for the presence of viruses. The organization accepts no liability
for any damage caused by any virus transmitted by this email.
=================================
