[ccp4bb]: Processing MAD data with radiation damage

[Tina Bakolitsa <bakolitsa@burnham.org>]

***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***

Hello all,

I am trying to phase using a 2-wavelength Se-MAD dataset (remote,
inflexion) collected in 10-degree wedges from a protein crystal that
suffers radiation damage. This damage can be seen clearly by a loss of
resolution after about 12 hours of exposure. I tried minimising exposure
time during collection (2 minutes/degree) but the symmetry is fairly low
(orthorhombic) and I don't have anything else (e.g. NCS, high solvent
content) that might help with phasing so tried to go for some redundancy
(>70% of reflections were measured 4 times or more). Crystal translation
wasn't really an option because of the special crystal orientation
requirements (bent loop) due to a long axis. The crystals are plates that
rarely grow single (<5% of the time) so I have a strong preference for the
1-crystal-1-structure MAD over the several-crystals MIR(AS) method (the
protein is >100 kDa so would most likely require several soaked-in
derivatives).

The resolution of the data collected was to 3.3 Å for the remote, with
SHARP indicating that the anomalous difference CC are not significant until
after 4.3 Å. I did not think that phasing by Se would be possible at this
resolution until the Science paper by Bass et al. came out in November
(v.298, p.1582) showing phasing by Se-MAD to 5 Å (but then they had 3
wavelengths, 7-fold NCS and high redundancy too).

The molecule/a.u. has 39 methionines. Patterson methods on their own
(SOLVE) have failed, giving low occupancy and high B factors which don't
refine to anything meaningful. Dual space methods (SHELXD) give 26
good-looking sites to 4.3 Å although only one is in density over 6 sigma.
The sites have low B-factors but gradually dwindling occupancies (from 1 to
0.4). I have checked by mass-spec and Se incorporation is 100%. The protein
is predicted to be all alpha-helical but even after solvent-flattening I
cannot see any sign of helical density.

Some statistics (pre solvent-flattening):

FOM (cen/acen)			0.398/0.606
Iso Phas. Power (inflexion)	1.073
Ano Phas. Power     (") 	1.410
Ano Phas. Power  (remote)	1.257

I'm currently looking at SHARP residual maps to see whether there is room
for improvment but my question is, are there any recommended methods for
dealing with such damaged data? For example, is it possible to process data
differentially depending on the stage/time at which they were collected?
Can adjustments be made for the decrease in the MAD signal? I took a
fluorescens scan at the beginning and end of data collection and saw a
significant decrease in the values of f' and f''. It is possible that the
crystal was not optimally oriented during the second scan - has anyone else
noticed this? If yes, is there a way to take this change into account when
looking for sites?

If not, i.e. if the consensus is that I should throw these data away and
try again, are there any other/newer methodologies recommended for data
collection and data processing involving radiation damage? Any suggestions
would be appreciated.

Thanks,

	Tina

----------------------------------------------------------------------
Tina Bakolitsa (PhD)

Lab 5107
The Burnham Institute			Tel: (858) 646 3100 ext. 3939
10901 North Torrey Pines Road		Fax: (858) 646 3196
La Jolla, CA 92037
-----------------------------------------------------------------------