I am refining a protein:substrate complex structure in which one
domain is showing very large movements such that electron density
virtually disappears at the greatest extents. In two separate
crystal forms, the domain in question is locked down by crystal
contacts, and I have very good data on the domain's fold. I figured
this would be a good time to use TLS refinement, separating the
domain as an individual unit. During my refmac5 restrained
refinement with TLS, I get nice decreases in R and Rfree along with
the dreaded:
Problem in MAKE_U_POSITIVE -3.9053693E-02 2.5330294E-02
special position ----------- 3.705357
special position ----------- 2.834380
special position ----------- 3.666230
An internet search located the following ccp4b thread:
http://www.dl.ac.uk/list-archive-public/ccp4bb/2006-01/msg00380.html
However, in that case the U was considerably more negative. I tried
using the TLSMD server to create potentially better domain
definitions (5 domains instead of 3), but I still get the warning. I
also tried increasing my starting B to 50 (based on Truncate output
and average B before TLS). Is this warning be something I should be
very worried about? Should I believe the output R and Rfree?
Jeffrey Wilson, Ph.D.
University of Cincinnati College of Medicine
Molecular Genetics Department
231 Albert Sabin Way
MSB 3109A
Cincinnati, OH 45267-0524
(513) 558-4651