Hi -
Single isomorphous replacement with anomalous scattering uses two
datasets that are isomorphous while using the anomalous signal in one
(here scattering and dispersion are synonymous?). How does SIRAS use
the anomalous scattering signal that's different than a single
wavelength anomalous dispersion ?
The anomalous signal in SAD or SIRAS is used in the same way - its the
difference from F+ and F-.
If you have on top a native the 'SIR' difference - isomorphous
differences - will be used as well
as a new source of information. "Weighting" these is rather
complicated but most software
handle that in ingenious and good ways - even if some can argue that
none of these ways is formally perfect.
How does an anomalous difference patterson (SAD) differ from an
isomorphous difference patterson with anomalous scattering (presumably
utilizing a native crystal)?
Depends what you:
If as 'SAD patterson' you define the 'anomalous patterson' this is
what it is. It the F+ vs F- patterson.
The isomorphous Patterson uses a different signal, the difference in
F(der) and F(native).
To get an "isomorphous difference patterson with anomalous scattering"
you would have to somehow
combine the two signals in some sort of "Fa". Programs like xprep can
do that to give you coefficients
to calculate such a Patterson, but it does not sound to me as a great
idea. I would calculate both
the 'anomalous' and 'isomorphous' Pattersons and compare.
Can a dataset that was not resolved by MAD/SAD be resolved by MIRAS/
SIRAS provided a good native dataset is available (Does MIRAS/SIRAS
offer more phasing power than a typical SAD dataset?)
The advantage of SAD/MAD is clearly that they are done on the same
crystal, so the isomorphism problem is non existent or in reality
minimal (some non-isomorphism especially in MAD might arise due to
radiation damage...).
If you do have a good native, that is isomorphous to your 'derivative'
its worth trying it out. Programs like SHARP, Solve/Phenix, CNS can
easily handle such scenarios. In the case of Se, having a SAD dataset
will give you about 6e signal per Se on the anomalous differences. The
Se-S "dispersive" differences are 18e, or three times as much. If you
would have a perfectly isomorphous native it the signal is triple ...
Something that you can - and very likely you should - do is try your
SAD map and your SIRAS map (if you have a 'derivative' and a 'native')
and look at them. In the past, I had often decided its useful to use
both since they have interestingly different features.
Hope these help - a good exercise in my phasing memory over (late)
morning coffee ...
Tassos
