[phenixbb] Multivariate Z score L-test

Claudia Scotti claudiascotti at hotmail.com
Fri Jun 29 02:04:44 PDT 2007

Dear List,

I'm only at my third structure (a scFv antibody fragment) and I'm disturbing about a dataset that is giving problems at the refinement stage: it gets stuck at about Rcryst= 0.36 and Rfree=0.44.

I've tried data processing with several softwares (Mosflm, XDS, Denzo) and the result is consistently P222, unit cell 69x79x90, 90, 90, 90. The data are anisotropic, with limits at 2.3 and 2.8 A. I cut the data at 3.0 A.
Matthews coefficient suggests two molecules per asu.
No NCS is detected by self-rotation and a native Patterson detects a small peak (10% of origin peak) at 0.19, 0.34, 0.5.
MR using several softwares (AmoRe, Molrep and Phaser) and several models all agree that the best space group is P21212, though the systematic absences are not perfectly consistent.

When refining, refinement gets stuck as I mentioned above.

I've run Phenix x-triage and the results include this line (see below): "It could be worthwhile considering reprocessing the data". I was wondering if the Multivariate Z score L-test: 6.305 might suggest how to reprocess the data.

Sorry for the probably trivial question and for probably missing important data.

Thanks a lot for any help and suggestions,


Twinning and intensity statistics summary (acentric data):

Statistics independent of twin laws
- /^2 : 1.827
- ^2/ : 0.838
-  : 0.649
- , : 0.417, 0.242
Multivariate Z score L-test: 6.305
The multivariate Z score is a quality measure of the given
spread in intensities. Good to reasonable data is expected
to have a Z score lower than 3.5.
Large values can indicate twinning, but small values do not
neccesarily exclude it.

No (pseudo)merohedral twin laws were found.

Patterson analyses
- Largest peak height : 11.094
(correpsonding p value : 1.080e-01)

The largest off-origin peak in the Patterson function is 11.09% of the
height of the origin peak. No significant pseudotranslation is detected.

The results of the L-test indicate that the intensity statistics
are significantly different then is expected from good to reasonable,
untwinned data.
As there are no twin laws possible given the crystal symmetry, there could be
a number of reasons for the departure of the intensity statistics from normality.
Overmerging pseudo-symmetric or twinned data, intensity to amplitude conversion problems
as well as bad data quality might be possible reasons.
It could be worthwhile considering reprocessing the data.

Claudia Scotti
Dipartimento di Medicina Sperimentale
Sezione di Patologia Generale
Universita' di Pavia
Piazza Botta, 10
27100 Pavia
Tel. 0039 0382 986335/8/1
Facs 0039 0382 303673

Express yourself instantly with MSN Messenger! Download today it's FREE!

More information about the phenixbb mailing list