Dear Dhirendra,

Your problems probably stems from trying to solve your structure using tantalum 
clusters at too high resolution. You may also have high anisotropy in your data 
since you get that negative eigenvalue error message in CTRUNCATE.

First, I suggest you look at the Harker sections from a Patterson synthesis up 
to say 6-7 Å resolution. If they look reasonable, solve that substructure 
(manually or with RSPS if you are rusty) and generate some phases to 6-7 Å from 
the anomalous data with your favourite program. Use these phases to look for 
free heavy atoms (originating from split clusters) using difference fouriers. 
The pseudo-translational symmetry may help you in analysing and solving the 
heavy-atom substructure. When you have included all significant sites you 
should then hopefully be able to see the four separate protomers. You can then 
easily generate vectors for averaging by searching with a cut-out sphere of 
density (you can use PHASER).  Make sure you see the pseudo-translational 
symmetry.

Then, phase extend (with for example DM) using the fourfold averaging in ~300 
steps starting from ~7-10 Å (try some variations to see what works best). I'd 
use 'O' to make and manually 3D-edit a good averaging mask (extend this mask to 
make a solvent mask, too). If the density isn't good enough after this you can 
try fourfold averaging in combination with multi-crystal averaging since your 
native was very non-isomorphous to your tantalum cluster data. Here, try phase 
extension using DMMULTI starting from ~7-10 Å in 300-500 steps (try some 
variations to see what works best).

In parallel, it wouldn't hurt to try some other (smaller) heavy atoms. You can 
use the phases from the tantalum cluster data to find them easily.

The CTRUNCATE problems you mention most probably stems from very high 
anisotropy in your data. We have found that the old truncate program works for 
some extremely anisotropic datasets (our typical datasets to be honest) when 
the new CTRUNCATE don't (giving the same error message as you got). Just select 
"old truncate" if you use the CCP4 interface. Later on, use the anisotropy 
server at NIH to analyse this. Using this server to generate an 
"anisotropically corrected" mtzfile will most probably enhance your results in 
the DM and DMMULTI sessions described above.

Cheers,

Martin

On Mar 6, 2010, at 12:52 AM, Dhirendra K Simanshu wrote:

> Dear all,
> 
> I am trying to phase 172 aa protein using tantalum bromide (Ta6Br12) cluster. 
> Data belongs to space group C2221 with 4 molecules (2 dimers) in the 
> asymmetric unit. Phenix.xtriage indicates presence of pseudo-translation 
> (peak height= 39). I have collected data at Ta peak (1.2550 A) to a 
> resolution of  2.9A, Ta inflection (1.2553 A) at 2.8 A using one crystal. I 
> also collected one high energy remote (1.059 A) data at 2.6 A using another 
> soaked crystal.  All three datasets shows presence of very good anomalous 
> signal till the highest resolution shell. 
> 
> Soaking with Tantalum bromide has changed the unit cell axes "a" and "b" by 
> 4-5 A so it is no longer isomorphous with the native data (2.7 A). But since 
> anomalous signal is very high in all these Ta soaked datasets, I have been 
> trying to find a solution using SAD/MAD but without any success. Processing 
> data in lower symmetry P21 and C2 also doesn't help.
> 
> Another noticeable thing is..... "scalepack2mtz" fails when I try to run to 
> convert my sca file into mtz (using CTRUNCATE option ON) with error saying 
> "Anisotropic correction failed - negative eigen value". 
> 
> 
> 
> Since finding a solution has not been possible for reasons unclear to me.  I 
> guess I am having problem because of presence of pseudo-translational 
> symmetry in my crystal. I would be grateful for any helpful advice or 
> suggestions in this regard. 
> 
> Thanks in advance!
> Regards
> Dhirendra
> 
> 
.
B. Martin Hallberg, PhD
Junior research fellow
Section for Cell Biology
Department of Cell and Molecular Biology
Karolinska Institutet
171 77 Stockholm
Sweden
http://tinyurl.com/yzn9y5j

Reply via email to