On top of what Eleanor mentioned, looking for about 160 S atoms with anomalous signal to 3.9A will be VERY hard. If you manage actually find those sites (or a good subset of them), the next density modification step can also be very hard: jumping over that 3.5-4A hurdle to get interpretable density right to your high reslution limit is often tricky. But ... with your high level of NCS you have a chance if you can get hold of some NCS operators to start with. For an example of 20-fold NCS and phasing to 6A see [1] (and GETAX is in CCP4, although without an interface in i2).
Do you get the feeling from your substructure solution trials that those sites are actually correct? Most of the stats at 3.9A (e.g. CCall/CCweak from SHELX) will not be very useful to judge the quality of a potential solution. What does the (relative) occupancy of all those sites look like? You don't want to see the first site with occ=1.0 and then a big jump with all other sites coming in at occ=0.3-0.4 or lower ... that wouldn't make a lot of sense. Also: do you see some kind of distinction between successful and unsuccecssful trials? If there is absolutely nothing that gives you a feeling for a successful substructure solution (even after trying with different resolution limits, NSIT parameters, programs etc), I would abandon that S-SAD approach and go for something with a bigger signal (and definitely think about more than just one wavelength if at all possible). What I often find to be crucial in such cases: really careful data processing (beamstop masking, correct (!) outlier rejection, ice-ring and damaged pixel handling etc). Regarding those operators from PROFESSS (I like that program a lot): they seem suspiciously close to 90/45/180 and 90/90/180, which always makes me suspicious. Not sure any of the above actually helps ... ;-) Cheers Clemens [1] http://scripts.iucr.org/cgi-bin/paper?S0907444902004614 On Sat, Jan 23, 2021 at 11:45:31AM +0000, Andrew Lovering wrote: > Dear All, > > > Having some fun with a marginal SAD S phasing case that you may be able to > help me on. > > My primary question is "does strong tNCS artefactually inflate FOM scores"? I > have a solution with FOM 0.48 to 3.5Ang or so and the map isn't as good as > I'd expect (in either hand). > > > -more detail if this helps: > > We have collected 3 sweeps at 1.77A wavelength, Rpim is good 2% or less, > xtriage Z-score is 2, aimless suggests anom signal to 3.9A, as does shelxc, > no signs of radiation damage, anom multiplicity is 40, cc anom 0.4 overall > > > Cell is P41212/P43212, 146.9 146.9 335.1 diffracts to 2Ang or so which > suggests 50% solvent = 22 copies in asu! > > > each copy should have 2 disulphides and 6 extra S in Met. I am using dsul > keyword in shelxd. I ask for a variety of sites, usually gets 240+, there is > also 300mM Mg and 20mM As in conditions (site completion in phaser with Mg > or As doesnt really help) > > > When we look at native patterson we see one major off origin peak at 0.5 0.5 > 0.173 fractional, 25% height of origin > > -and a smaller peak at 0.35 0.35 0.5 10% of origin > > > I have tried everything. Changing phasing limits, programme, cutting back > data range, number of sites. But I can get a decent sites pdb from > shelx/crank2 which popped into phenix autosol gives stats that wrongly make > me think I am close (fom 0.4-0.5, score 30+, skew 0.1+). My model has no > molrep possibilities and likely no alpha helical content > > > If I run the substructure sites through profess, I get two nice operators: > > -(op1) polar 90 -44 179.9 trans -72 -73 -444.6 rotation order 8, N atoms > paired 74 from 16 of input atoms > > -(op2) polar 90 -87 180 trans -147 -9 -863 rotation order 2 paired 8 atoms > from 8 of input > > (profess notes these two sit 43 degrees apart) > > > I can see that op1 has a translational component that is my tNCS vector (-0.5 > -0.5 and 0.17 away from 0.5)....and that likely I have 4 fold operator that > sits over my crystallographic 2-fold, which is 45 degree from another 2 fold > (self rotation function attached..). I would welcome a better description of > this 😊 > > Using these operators (or indeed a pure tNCS only operator) in dm gives good > correlation but again no map improvement. Is it worth playing with nmol? > > > Any suggestions welcome, before we hit this with some higher electron > derivatives! > > > Thanks > > Andy > > > > > > ________________________________ > From: > > Sent: January 23, 2021 11:08 AM > To: Andrew Lovering (Biosciences) > Subject: Srf > > > ######################################################################## > > 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/ -- *-------------------------------------------------------------- * Clemens Vonrhein, Ph.D. vonrhein AT GlobalPhasing DOT com * Global Phasing Ltd., Sheraton House, Castle Park * Cambridge CB3 0AX, UK www.globalphasing.com *-------------------------------------------------------------- ######################################################################## 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/
