Lachlan Cranswick wrote :
> Given ESPOIR uses extracted intensities - has it been
> tested/how good is it on single crystal HKL data
> (0% overlap) in suggesting a starting model?
> Am mainly thinking in terms of datasets with poor intensity,
> ambigious spacegroup, etc - but it should work by default
> on good quality data(?).
>> Surely if ESPOIR/Monte Carlo methods can do a good job with
> intensities extracted from powder data, it should be able to do
> at least as well on single crystal data(?)
In general, who can do the hard work, can do the simple.
Single crystal data-like were tested.
Yes, ESPOIR is able to determine structures from single crystal
data. But the upper complexity limit is ridiculously low if
compared with direct methods. In the "from scratch" option
(i.e. starting from completely random models), the maximum
number of independent atoms is 15-30 : really small ! The
maximum complexity for using ESPOIR in the MR (Molecular
Replacement) mode (locating a known fragment) is not well
known, but degrees of freedom (DoF) may be of the order 50-100
(however, the ESPOIR 2.00 version can locate only one
fragment, corresponding to 6 DoF). Further developments will
concern the consideration of torsion angles and rotating/translating
several fragments together.
Reasons for not being able to determine a structure from
single crystal data may be quite numerous. If the crystal is
twinned by merohedry, ESPOIR will be very helpful, for instance,
since an option for working from a powder pattern regenerated from
the |Fs| is included. This option will restitute a non-twinned set of
powder data, though loosing some information in strict orverlapping
Armel Le Bail - Université du Maine, Laboratoire des Fluorures,
CNRS ESA 6010, Av. O. Messiaen, 72085 Le Mans Cedex 9, France