On Fri, August 17, 2007 10:56 am, Eleanor Dodson wrote:
> There are obvious ones like - incomplete structure etc, but have you
> tried TLS? Sometimes this can dramaticaly improve the R factors.
>
When people run TLS, are they trying to improve the R-factors, or actually
generate a reasonable model for the structure? I really think people
should be extremely cautious in just turning on the TLS option with dozens
of arbitrary groups during refinement without thinking about the resultant
model.
Bernie Santarsiero
|