Dear all,
coming back to those map questions again:
On Wed, Nov 25, 2020 at 12:56:16PM +0000, Mooers, Blaine H.M. (HSC) wrote:
> The question is about a Fo-Fc map and the replies have been focused on such
> maps.
> However, I question why you are bothering with Fo-Fc maps.
> The ligand was soaked into the crystal. There is probably an isomorphous apo
> data set available.
> A [Fo(ligand complex) - Fo(apo)]*exp(alpha_apo,calc) map should be consulted
> in
> preference to various Fo-Fc maps corrected for phase bias.
> It would be even better to substitute in the experimental phases for the apo
> structure if
> they are available.
>
> Fo-Fo maps may be nosier than Fo-Fc maps. but they are more reliable when you
> are
> trying to decide if a ligand is present.
Generally I would agree, but with a slight caveat: the accurate
scaling of those two sets of Fo can become tricky. If the Fo(apo) was
collected in a very different way (different beamline, different
detector, different protocol, different processing package, re-using a
PDB deposition from 10 years ago etc) compared to Fo(soak), things can
become difficult: the required scaling of noisy measurements to extract
the very small ligand signal has to be very accurate.
So when doing high- or low-throughput ligand screening data
collection, maybe we shoulld always collect another APO dataset during
teh same beamline shift? Of course, this then start to play the same
role as the "ground-state" map in PanDDA (if I understood this
correctly) ... but a very isomorphous (in terms of experiment,
hardware, processing) experimental APO dataset seems like a good idea
nevertheless.
> I second Robbie's "bloody obvious" rule.
Ditto.
I like to start with the null hypothesis that the experiment (soaking
a crystal with a compound solution) has not worked (ligand has not
bound). That seems a better "bias" than the other way round ("I soaked
therefore it is there.").
Our ligand-detection maps approach three questions in turn - each
depending on a positive outcome of the previous:
(1) Has something bound (=> detection of residual difference density
under the null hypothesis)?
(2) Where has it bound (=> detection of what we call "interesting
regions")?
(3) How has it bound (=> removal or potentially wrong dummy
water/atoms to result in shape description of binding excluding
bulk-solvent)
So instead of zooming in on our favourite binding site (which fits so
nicely that beautiful theory I have) and immediately dialing map
levels down or playing with B-factor sharpening, it forces us to take
several steps back from any potential bias I think. Note that map
levels and B-factor sharpening/blurring are very useful tools that can
become crucial in the correct interpretation of compound density, but
they tend to mainly modify shape and levels while easily pushing the
initial question ("Has this small compound bound?") into the background.
Anyway, just another minor thoughts ...
Cheers
Clemens
https://www.globalphasing.com/buster/wiki/index.cgi?DifferenceFourierMaps#iso
https://www.globalphasing.com/buster/wiki/index.cgi?LigandDetectionModes
########################################################################
To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1
This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a mailing list
hosted by www.jiscmail.ac.uk, terms & conditions are available at
https://www.jiscmail.ac.uk/policyandsecurity/
