Hi Kai A few things to reiterate:
One: you have to collect the data VERY carefully. Many beamlines have by now tried to make this easier or even transparent, but it's impossible to make it foolproof. The signal is very weak, and it is really easy to degrade. - Radiation damage - that's the killer (as Mathew Snee described) - Over-exposing the low resolution reflections - that was a stupid-easy thing to do on CCD detectors, and even easier before we collectively realised that it's not resolution we should chase. - Lots of other stuff. Read James Holton's review from 2009 (https://journals.iucr.org/s/issues/2009/02/00/xh5015/index.html). The key is that the dataset is as absolutely uniform as possible. You only need to phase a small fraction of the reflections - I remember the late George Sheldrick saying <5% is plenty - but those have to be evenly spread through reciprocal space. Two: Be ruthless in throwing away bad images. It's more important to NOT have bad images than to HAVE a bit more dodgy signal. If you're unlucky, you may have to try multiple combinations of good images - and you need to scale and merge and then analyse each combination separately, because you can only know if it worked by trying to find the substructure. Three: That one-atom solution you describe sounds like the "Uranium atom" trivial and false solution that direct methods readily converge on (including the shake-and-bake that ShelxE uses). (Something like that - I'm parroting George from almost 3 decades ago). Not it MAY be correct, as Kay suggests below - but back when I was doing this a lot, when I saw it I just stopped looking and went back upstream: cut the data differently (again), or indeed just discard the data altogether and try a new dataset or in fact crystal. Good luck! Frank -----Original Message----- From: CCP4 bulletin board <[email protected]> On Behalf Of Kay Diederichs Sent: 11 July 2026 21:54 To: [email protected] Subject: Re: [ccp4bb] SAD Hi Kai, what do you call "very weak anomalous signal"? To understand the situation, we need numbers, e.g. CC1/2anom as a function of resolution. In my experience, a value above 50% for data in the lowest resolution range is hopeful. At which resolution does CC1/2anom go below 30% ? You don't say whether the CCall/CCweak distribution in SHELXD is bimodal. If it is, the substructure is very likely solved and useful. There may be nothing wrong with your occupancy pattern. SHELXD considers all anomalous scatterers. You may have one metal with high f", producing the first site, and 16 others representing your SeMet sites. You don't say what SHELXE gives - if the two solutions (one of them with the -i option) differ, then the structure's hand is established. And finally, proper data processing is essential. Do not rely on automatic processing from a pipeline running at the synchrotron. Radiation damage must be assessed as a function of frame number, and the useful range established. Zero-dose extrapolation may be an option. Hope this helps - if you have questions, send me a private email. best, Kay On Sat, 11 Jul 2026 11:40:56 +0000, [email protected] <[email protected]> wrote: >Dear all, >I am currently processing a SAD dataset collected from a Se-Met derivative. >According to the matthews_coef calculation in CCP4, there should be 20 heavy >atom sites in the asymmetric unit. >However, due to the very weak anomalous signal, the occupancy refinement in >SHELX shows only one site with an occupancy above 1.0, one site around 0.4, >and the remaining 15 sites all converging to approximately 0.3. >Given this situation, I would be very grateful for any suggestions on how to >obtain reliable initial phases. Are there any specific strategies, software >settings, or alternative approaches (e.g., density modification or partial >structure refinement) that you would recommend to make the most of this weak >signal? >Thank you in advance for your time and insights. >Best regards, >Kai Zhang > > >######################################################################## > >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/ ######################################################################## 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/ ######################################################################## 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/
