[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[ccp4bb]: Summary: Molecular Replacement woes!



***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***

hi everyone,
thankyou very much for the suggestions. there were many, therefore i will try 
my best to keep it brief.

Louis Lim:
If you started your minimization with the full high resolution data set, try 
limiting the resolution and do some rigid body refinement, gradually extending 
the resolution. Then you may get to a better starting point for 
minimization/dynamics/etc. The MR solution you described seems very clear and 
is most likely correct.

Frank Vondelft:
Orthorhombic space groups that are NOT P212121 are really very rare, so it is 
very likely that the wrong spacegroup has been assigned. 
If the assignment was based on the systematic absences printed by scalepack it 
is possible that if one never actually recorded the relevant reflections to 
begin with, scalepack doesn't print them either - and doesn't tell you that 
they're not measured. But in P222 they're easy to find, so scale in P222 and 
then look for them in the .sca file.
other programs one could try: amore, molrep, epmr, beast.
Fred Vellieux:
do check carefully the systematic absences along a*, b*, c* to confirm that 
you have the correct space group and you may have to carry out translational 
searches in all primitive orthorhombic space groups.

David Mandelman:
Based on your cross rot. results, it is possible that you have 2 mol.'s in the 
asymmetric unit (A.U.) run thru a quick Matthews calc (Vm) to figure out the 
number of mol.'s in the A.U. To do this:
                          Vm = (A*B*C)/(4*M.W.*Z)
A,B,C are your unit cell lengths
4 is the number of symm op.'s for the orthorhombic space group
M.W. is the mol. weight of your molecule
and Z is the unknown number of mol.'s in the A.U.
You supply values of Z until your Vm falls between 1.7 - 3.5, a typical value 
for most proteins. Of course, not all prots will fall in this range, but if 
you find a value of 2, then you may want to look for a second molecule in the 
A.U., you already have the rot. sol'n. for it.

Anastassis Perrakis:
"click at ccp4i the button 'Cell contents nalysis' and give the
mtz file name and the MW of the protein"
    In ccp4i, when you run molecular replacement programs there is a button 
provided for the case that the 'space group hypothesis' is wrong. One can use 
ccp4i very easily to try all P-orhtorhombic hypotheses,  That might be usefull 
especially in cases that one of the axes of the crystal was perfectly aligned 
with the spindle in data collection and you miss all h00/0k0/00l reflections 
and can not decide really if there is a screw axis or not.
    Run arp/warp on molecular replacement solutions if data are in a decent 
resolution: 2.0 A or better for autobuilding, or 2.0-3.0 for just 'model 
update'
combined with refmac refinement. And 'resolve' for 'prime and switch' - it 
tends to give nice bias-reduced maps from molrep solutions.

Jim Pflugrath:
"I believe these crystals have the a and b cell lengths of almost the same 
length. I believe there might be a possibility that these crystals are 
twinned. Possibly by a 90 degree rotation around the c axis."

Leif Hanson:
Nucleosomes have crystallized as P 2(1) and P 2(1)2(1)2(1) in the past. 
Unlikely that that the space group assignment is correct. Search with the 
histones or with the DNA but don't do both initially. Use only the core 
histones, no tails. You may get two solutions at this point. Look at a map and 
see if you can find the DNA dyad. The DNA is especially well defined here. If 
you scale the map properly you should see the P atoms well without much 
confusing density.

Wrong space group asssignment being almost unanimously suggested i did rethink 
the issue. Matthew's calculation suggests one molecule per AU. I looked at the 
data again and confirmed the existence of only one screw axis. it is 
definitely not p212121. but i am not so sure it is orthorhombic anymore. i 
dont have any arguments against the possibility of monoclinic. i have indexed 
and scaled the data in p21 and as expected the completion suffers because we 
assumed it was orthorhombic for the data collection strategy. Havent had much 
success with molecular replacement yet. 
I will try some of the other MR softwares suggested.
I havent included everyone's suggestions in the summary but I sincerely thank 
everyone for your input.
cheenu.


Srinivas Chakravarthy
Colorado State University
Fort Collins
CO-80521