reloris at vub.ac.be
Sat Feb 10 07:05:44 EST 1996
For some time we are trying to refine a mutant of RNase T1.
The data are of reasonable quality (rmerge 8%, redundancy about 5,
resolution 2.3 A amd completeness 98%). The starting model is an
isomorphous model of another mutant refined at 1.8 A.
THe starting Rvalue and Rfree is after rigid body refinement 0.36
At this point, positional and B-value refinement were attempted,
but no drop in Rfree could be obtained. Therefore, we inspected the
electron density map, which turned out to be very clear. An inhibitor
molecule (2'GMP) and 54 water molecules were added to the model and one side
chain that had no density was deleted. Also the mutation was modelled
(which is Tyr - > Ala and thus is no more than deleting the
appropriate side chain). No other errors could be found in the model.
THis drops both R and Rfree equally. But again, still no positional
refinement seems to be possible. B-value refinement now drops
Rfree with half the amount that R drops.
What is goiing on here? Although the starting model comes from
higher resolution data than the structure that we try to refine,
I would expect that at leasat with a good 2.3 A dataset some
sensible refinement can be done! I would expect such a behavier
for let say 3A data, but to se it here is somewhat a surprise to me.
Any suggestion about what to do. If the final conclusion is indeed
that refinement should not be done since the original model is the
best possible representation of this structure, how does one convince
referees that this is so?
(reloris at vub.ac.be)
More information about the X-plor