The dose rate at ALS 5.0.2 should be about 100 kGy/s with standard
settings. Lindsey? What was your exposure time? Was the crystal
bigger than the beam? You need these two bits of information to do a
rough estimate of dose per image. Without this information any strategy
is just a wild guess.
I always recommend inverse beam and round-robin multi-wavelength MAD as
the first thing to try. This is because inverse beam and interleaved
wavelengths minimize the impact of radiation damage on the differences
you are trying to measure. Yes, it may be possible to solve a structure
with a quick-and-dirty SAD spin, but if you only have one good crystal I
advise strongly against such a gamble. Better to do the data collection
in a way that maximizes your chances of success. This depends on the
equipment you are using! Not every beamline is the same. In this case
ALS 5.0.2 has no issues with inverse beam or fast wavelength changes.
In fact, all beamlines in California (ALS and SSRL) are optimized for
the inverse-beam, round-robin MAD strategy. Other beamlines may have
optimized something else. Always ask your local beamline scientist!
The only thing that could have been "wrong" with your collection
strategy is the exposure time. Too long and you probably do have
radiation damage problems. Even if you do, however, it may still be
possible to solve your structure. I have never seen a SeMet site fade
faster than 5 MGy (half-dose), so your first 50 seconds of data
collection on ALS 5.0.2 should not have significant damage. Unless, of
course, you have set a new world record for sensitive SeMet. In that
case, you might want to try RIP phasing. For RIP phasing you treat your
early data and late data as two separate data sets: derivative and
native, respectively.
I don't know why so many people assume that inverse beam and
multi-wavelengths means you are over-exposing your crystal. This is only
true if you don't update your exposure time. The number of photons your
crystal will give off before it dies is fixed. All you get to do is
decide how many images you want to use to store those photons. If you
double the number of images you plan to collect (such as introducing
inverse beam) then you should cut your exposure time in half. If you
quadruple the number of images (such as two wavelengths with inverse
beam), then divide your exposure by four. etc. It is always OK to keep
collecting beyond your estimated "damage limit" just in case you were
being too pessimistic, but no matter how you collect your data you
should be prepared to throw out some of the last images.
Anyway, Lindsey, if you could provide your exposure time, and perhaps
your crystal's rough size, I think you will get a lot better suggestions
about what to try next.
-James Holton
MAD Scientist
On 8/27/2019 11:14 AM, Nukri Sanishvili wrote:
Hi Lindsey,
As I mentioned to you in the separate email, 180 degrees for each half
is too little.
Here I'll try to explain some more about SAD vs. MAD:
What I have observed at our beamlines is that the majority of those
who collect MAD data, do it as as an afterthought of SAD. Priority in
these experiments is given to SAD and after it is done, some folks
decide to collect more data at a different wavelength "just in case".
There are two big mistakes in this approach, also explaining why SAD
so often "works better than MAD":
1. Unless one collects the second wavelength from a fresh part of the
crystal, or from a different crystal, there is too much radiation
damage in the second data set. Therefore, the differences in the
intensities are mostly caused by radiation damage and not by anomalous
or dispersive signal. This, of course, kills phasing. This is how SAD
can be "better" than MAD.
In a proper experiment, both wavelengths must be given equal priority.
I.e. distribute the crystal life time equally between the two.
2. Another mistake also stems from the fact that the 2nd wavelength is
treated as "addition" to SAD. Whether it is optimal or not is a
different discussion but typically, the SAD data are collected at the
absorption peak. Then, for 2-wavelength, one collects inflection
point. What is lost in this approach is the whole purpose of a MAD
experiment, which is to use the dispersive signal along with the
anomalous one. Dispersive signal between the inflection point and the
peak wavelengths is not so great. In a good experiment, one of the
wavelengths is at the inflection point (as a must). One could argue
that the other is not at the absorption peak but above the peak (in
energy) in order to increase the dispersive signal. How much above,
will depend on particular f' and f" plots. Further the better for the
dispersive signal, but you also want to retain good anomalous one. So,
some compromise needs to be made here.
Bottom line is that SAD and MAD are two different experiments and one
is not a simple expansion of the other. You need to make a decision
which one you are doing and collect data accordingly.
Cheers,
Nukri
On Tue, Aug 27, 2019 at 11:30 AM Doyle, Lindsey A
<[email protected] <mailto:[email protected]>> wrote:
Hi Nukri,
Thanks so much for your response. I appreciate the advice.
1. Yes, I verified that the anomalous option is turned on during
data processing. Always a good question to ask
2. I collected 180° for each half. I have not tried phasing just
one half. I’ll give a try but with my space group being P 21 the
completeness and redundancy might be pretty low. I have a couple
inverse beam data sets with wedges of 5° but they were about the
same as the ones with 1°
3. I’ve been collecting 0.5 sec exposures but without reducing the
flux. This seems to be one of the most recommended things and will
be definitely doing it on my next run.
4. I’ve tried both SAD and MAD with peak 12661 (0.9793 Å) and
inflection 12658 (0.9795 Å)
Thanks again,
Lindsey
On Aug 26, 2019, at 6:54 PM, Nukri Sanishvili <[email protected]
<mailto:[email protected]>> wrote:
Hi Lindsey,
Obviously, one would need a lot more information to properly
diagnose the problem and I am sure much smarter people them me
will ask you for that. But just to move the task by couple of
steps, I want to point out couple of things.
1. Trivial question: did you have the anomalous option turned on
during data processing? (Just like from the IT help - is your
computer turned on?)
2. How much data did you collect for each half of the inverse
beam geometry? If you have enough, try phasing with only one
half. When done properly, inverse beam experiment is great but it
can easily get tricky introducing systematic errors and thus
swamping anomalous signal. If you redo the inverse beam, use
little wider wedges, say, 5-10 degrees.
3. I thought an example of diffraction image would not give any
useful information but... Judging by how smooth the background is
on your Pilatus image, I am guessing you have used a lot of
exposure. Can you calculate how much dose did you put in your
crystal? If you are going to re-do the experiment, I would
suggest reducing the exposure level and collecting more data.
4. Because you are not showing f' and f" plots, I am guessing
that you are doing SAD. If it fails and you end up redoing your
experiment and you have crystals for it, you might want to try
2-wavelength MAD but for that you would need to know exactly
where is inflection point and collect one of the datasets there.
Good luck!
Nukri
On Mon, Aug 26, 2019 at 5:45 PM L. Doyle <[email protected]
<mailto:[email protected]>> wrote:
I have some Seleno-Methionine protein crystals (12 SeMet of
211 amino acids, incorporation verified by Mass Spec). I've
already collected several datasets (ALS BL5.0.2) but I seem
to be losing (rejecting?) a lot of anomalous signal during
data processing. I'm most familiar with HKL2000, but I have
tried XDS and DIALS auto-processing. Here is a scan:
https://ibb.co/LZqm33p
<https://urldefense.proofpoint.com/v2/url?u=https-3A__ibb.co_LZqm33p&d=DwMFaQ&c=eRAMFD45gAfqt84VtBcfhQ&r=2fgF7nXnZhu6kQ1ZLLvJeA&m=0-kIFi_aWaNeQ9Rl0Ca1A5Z5P41UhmeHF5atpPFQ1x8&s=AYn3f5GZQ2ZQv5oq-gYN2HEbikhDEnKryiUsbaQevVE&e=>
and here is an example of a frame: https://ibb.co/gR3ZR47
<https://urldefense.proofpoint.com/v2/url?u=https-3A__ibb.co_gR3ZR47&d=DwMFaQ&c=eRAMFD45gAfqt84VtBcfhQ&r=2fgF7nXnZhu6kQ1ZLLvJeA&m=0-kIFi_aWaNeQ9Rl0Ca1A5Z5P41UhmeHF5atpPFQ1x8&s=MmgiUiUT2TmSx0xx3RskenKPWOJfYoTnLKXo2P8x2mw&e=>.
Each frame is 0.25° and I'm using inverse beam with wedge
size 1°. Maybe I need to adjust my collection strategy? All
previous datasets have been in space group P 21 with
dimensions of approx. 24.5Å, 85Å, 40Å, 90°, 96.5°, 90°. I'm
sure there are additional things I can be doing in HKL but
I've run out of ideas. Any advice or recommendations would be
appreciated. Please let me know if you need additional
information.
Thank you,
Lindsey
########################################################################
To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB&A=1
<https://urldefense.proofpoint.com/v2/url?u=https-3A__www.jiscmail.ac.uk_cgi-2Dbin_webadmin-3FSUBED1-3DCCP4BB-26A-3D1&d=DwMFaQ&c=eRAMFD45gAfqt84VtBcfhQ&r=2fgF7nXnZhu6kQ1ZLLvJeA&m=0-kIFi_aWaNeQ9Rl0Ca1A5Z5P41UhmeHF5atpPFQ1x8&s=SBBWJJiu3emknBfFcxVTI545aZP1sEHSkBcmSVcwtqM&e=>
------------------------------------------------------------------------
To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB&A=1
########################################################################
To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCP4BB&A=1
