[phenixbb] bug?

Patrick Loll pat.loll at drexel.edu
Thu May 12 14:02:07 PDT 2016

I just ran into something that might be a small bug (phenix 1.10.1-2155)

Background: I have a problem data set for which the unit cell is too small to hold we thought we were crystallizing (which probably means we crystallized the wrong protein). In any case, I ran the data through Xtriage, supplying the sequence file for the too-large protein, even though I knew this is probably not correct. This sequence file contains 264 residues.  

I was surprised to see that the program returned with a perfectly reasonable estimate of the Matthews coefficient, making me briefly hopeful that I had been doing the calculation incorrectly. Alas, not so; when I dig through the log file, I find the following:

____begin excerpt_______________________________________________________________

 determining composition from sequence file /xxxx/seq.dat                     
Too many residues to fit in the ASU
 resetting numer of residues in monomer to    26

=================== Solvent content and Matthews coefficient ==================

Crystallized molecule(s) defined as 264 protein residues
 | Solvent content analysis                                        |
 | Copies | Solvent content | Matthews coeff. | P(solvent content) |
 | 1      | 0.666           | 3.68            | 0.621              |
 | 2      | 0.331           | 1.84            | 0.379              |

Best guess : 1 copy in the ASU

_____end excerpt_______________________________________________________________

So it seems as though the program, when confronted with a protein mass that’s too large for the AU, divides the mass by 10 and tries again. However, this is not clear from the results; in the results section, it states “Crystallized molecule(s) defined as 264 protein residues." So a user might be misled into thinking that everything is OK, when in fact it’s not…


Patrick J. Loll, Ph. D.  
Professor of Biochemistry & Molecular Biology
Director, Biochemistry Graduate Program
Drexel University College of Medicine
Room 10-102 New College Building
245 N. 15th St., Mailstop 497
Philadelphia, PA  19102-1192  USA

(215) 762-7706
pat.loll at drexelmed.edu

More information about the phenixbb mailing list