Thanks to everyone who helped with the self RF problem: Eleanor, Ian,  
Claudine, Pietro & Alexei.

Eleanor wrote:

> 1) It is a bit hard to find out how MOLREP defines its orthogonal  
> axes - many programs use X0 || a, Yo || b* and in P21 hence Zortho  
> is || to c*
> If that is what Molrep does then your 2 fold is in the a c* plane,  
> 21 degrees or 111 degrees from c*.
> The 2 peaks you see are symmetry equivalents.

This was my interpretation. Glad we agree ;-) The documentation says  
"A parallel to X , Cstar parallel to Z"

> As  for the Patterson - what height are those peaks relative to the  
> origin?

The peaks are  u = 0.129, v = 0.473, w = 0.220 (20% of origin peak  
height) and u = 0.180, v = 0.500, w = 0.248 (19%). What I don't get is  
why there are two and only one strong 2-fold. 2 dimers in the AU gives  
50% solvent, 1 dimer 75%. The crystals diffract to 2.3Å, which would  
tip the balance in favour of 50% solvent in my opinion.

> With 2 dimers in the asymm unit and with the non-cryst 2-fold  
> perpendicular to b* you could have such translations between one  
> monomer and another.

Would the 2-folds of both dimers have to be very similarly oriented?  
Maybe one peak masks the other at this resolution?

> is there a model - easiest to solve it then analyse this sort of  
> stuff later!

Believe me, we've been trying for a very long time! The problem is  
that it's a leucine rich repeat protein with under 30% sequence  
identity to any of the other LRR models out there. I think the failure  
of MR is down to a combination of a) the low homology, b) the  
pseudosymmetry, c) the nature of the LRR, which means you can get MR  
solutions that are out by one or more repeats. Maybe even the internal  
symmetry of the whole LRR structure can add to this pathology? We've  
had some solutions that looked almost right, but we can never see much  
more than what's already in the MR solution.

Ian wrote:

> The symmetry of the self-RF is explained in detail in the  
> documentation for POLARRFN, in fact I would advise you to use this  
> because you can then plot monoclinic space groups with the unique b  
> axis along the orthogonal Z axis (NCODE = 3) and then the symmetry  
> is *much* easier to interpret.

The reason I started using Molrep was that POLARRFN always used to  
choke on these data. However that problem seems to have disappeared.  
Using ORTH 3 indeed gives a more interpretable plot, as you say.

> According to polarrfn.doc the symmetry generated by a 2-fold along b  
> parallel to Z is (180-theta, 180-phi, kappa) so the peak in the list  
> (159,180,180) is the same as (21,0,180) which is a NCS 2-fold that  
> you can see just below centre.  The peak (111,0,180) is thus the  
> same as (69,180,180) near the top which is another NCS 2-fold perp  
> to the first generated by the crystallographic 2-fold.

Indeed, I see the peak (69, 180, 180) but I don't find it in the list  
in the log file from Molrep. I thought that list was supposed to be  
exhaustive. Also the plot is not well documented for Molrep. I wrote  
to the BB a while ago to ask what the contour levels were but no-one  
answered. By Googling I found a crystallisation paper where it was  
described as "from 0.5 sigma in steps of 0.5 sigma" but that  
information appears to have come by word of mouth. Also, is it just  
the "north hemisphere", as Claudine put it, that is plotted?

Anyway, I feel somewhat wiser now...

Derek