[ccp4bb] I/sigma for h+k=2n+1 and h+k=2n
Hi, I had a dataset which is P21 but with a pseudo-translational symmetry of (1/2, 1/2 ,0). Theoretically the dataset should show systematic weak spots of h+K= 2n+1 compared to h+k= 2n. Is that correct? I would like to have a close look at the reflections. for example, the average I/sigma for reflections with h+k=2n+1 and reflections with h+k=2n. Which software could do this job? A brief tutorial is appreciated. Yu -- Yu Zhang HHMI associate Waksman Institute, Rutgers University 190 Frelinghuysen Rd. Piscataway, NJ, 08904
Re: [ccp4bb] I/sigma for h+k=2n+1 and h+k=2n
On Tue, 2011-07-26 at 09:51 -0400, zhang yu wrote: > I would like to have a close look at the reflections. for example, the > average I/sigma for reflections with h+k=2n+1 and reflections with h > +k=2n. Once you have your reflections listed in a file having h,k,l,f,sigf as first five columns (e.g. scalepack output), this trivial one-liner will print even (h+k) reflections awk '{if(($1+$2)%2==0) print;}' foo.sca It's easy to get the reflections in text format out of an mtz file: mtzdmp foo.mtz -n -1 With that said, I suspect xtriage may be helpful in identifying pseudotranslational NCS, as well as good ole Patterson synthesis. Cheers, Ed. -- Oh, suddenly throwing a giraffe into a volcano to make water is crazy? Julian, King of Lemurs
Re: [ccp4bb] I/sigma for h+k=2n+1 and h+k=2n
Yu, There is a parity analysis in dataman (a USF program) for example. You have to take into account that the sigmas are generally estimated assuming a unimodal intensity distribution, which is no longer true in the pseudo-symmetric case. In practice this means that the sigmas of the strong reflections tend to be underestimated (generally not a problem really) and those of the weak reflections are overestimated. This can be avoided to some extent by scaling the h+k = 2n and h+k = 2n +1 reflections separately. I ended up writing a small python script to do this from XDS output and scaling separately (see Oksanen et al. (2006) Acta Cryst. D62 1369-1374). Of course it would be even better if the scaling program would directly take into account the bimodal distribution... HTH, Esko On 26.7.2011, at 15.51, zhang yu wrote: Hi, I had a dataset which is P21 but with a pseudo-translational symmetry of (1/2, 1/2 ,0). Theoretically the dataset should show systematic weak spots of h+K= 2n+1 compared to h+k= 2n. Is that correct? I would like to have a close look at the reflections. for example, the average I/sigma for reflections with h+k=2n+1 and reflections with h+k=2n. Which software could do this job? A brief tutorial is appreciated. Yu -- Yu Zhang HHMI associate Waksman Institute, Rutgers University 190 Frelinghuysen Rd. Piscataway, NJ, 08904 Esko Oksanen, PhD Post-doctoral Fellow (EMBO) Groupe Synchrotron, Institut de Biologie Structurale J.P. Ebel 41, rue Jules Horowitz F-38027 GRENOBLE Cedex 1 FRANCE tel. +33 4 38 78 95 96 mob. +33 6 84 15 14 88
Re: [ccp4bb] I/sigma for h+k=2n+1 and h+k=2n
Hi Thanks for all the replies. I ran TRUNCATE in CCP4 and got what I want. I will try other options later. Thank you again. Yu 2011/7/26 Esko Oksanen > Yu, > > There is a parity analysis in dataman (a USF program) for example. You > have to take into account that the sigmas are generally estimated assuming a > unimodal intensity distribution, which is no longer true in the > pseudo-symmetric case. In practice this means that the sigmas of the strong > reflections tend to be underestimated (generally not a problem really) and > those of the weak reflections are overestimated. This can be avoided to some > extent by scaling the h+k = 2n and h+k = 2n+1 reflections separately. I > ended up writing a small python script to do this from XDS output and > scaling separately (see Oksanen et al. (2006) Acta Cryst. D62 1369-1374). Of > course it would be even better if the scaling program would directly take > into account the bimodal distribution... > > HTH, > Esko > > > On 26.7.2011, at 15.51, zhang yu wrote: > > Hi, >> >> I had a dataset which is P21 but with a pseudo-translational symmetry of >> (1/2, 1/2 ,0). Theoretically the dataset should show systematic weak spots >> of h+K= 2n+1 compared to h+k= 2n. Is that correct? >> >> I would like to have a close look at the reflections. for example, the >> average I/sigma for reflections with h+k=2n+1 and reflections with h+k=2n. >> Which software could do this job? A brief tutorial is appreciated. >> >> Yu >> >> >> -- >> Yu Zhang >> HHMI associate >> Waksman Institute, Rutgers University >> 190 Frelinghuysen Rd. >> Piscataway, NJ, 08904 >> >> >> > > > Esko Oksanen, PhD > Post-doctoral Fellow (EMBO) > Groupe Synchrotron, Institut de Biologie Structurale J.P. Ebel > 41, rue Jules Horowitz > F-38027 GRENOBLE Cedex 1 > FRANCE > tel. +33 4 38 78 95 96 > mob. +33 6 84 15 14 88 > > > > > > > > -- Yu Zhang HHMI associate Waksman Institute, Rutgers University 190 Frelinghuysen Rd. Piscataway, NJ, 08904
Re: [ccp4bb] I/sigma for h+k=2n+1 and h+k=2n
>From memory - check the documentation mtzutils hklin1 data.mtz RZONE 1 1 0 2 0 lists reflections with h+k = 2n END mtzutils hklin1 data.mtz RZONE 1 1 0 2 1 lists reflections with h+k = 2n+1 END Eleanor This will list all On Tue, 26 Jul 2011 09:51:47 -0400, zhang yu wrote: > Hi, > > I had a dataset which is P21 but with a pseudo-translational symmetry of > (1/2, 1/2 ,0). Theoretically the dataset should show systematic weak spots > of h+K= 2n+1 compared to h+k= 2n. Is that correct? > > I would like to have a close look at the reflections. for example, the > average I/sigma for reflections with h+k=2n+1 and reflections with h+k=2n. > Which software could do this job? A brief tutorial is appreciated. > > Yu