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

No Subject



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

Thanks to all that replied to the following question.
All suggestions are found at the end of this email.

> I am refining a structure at 1.9 A resolution, spacegroup P212121, one
> molecule per assymetric unit. Unfortunatelly the primary structure for this
> protein is unkonw (purified protein), but its tertiary structure was solved
> (main chain and pseudo-side chains). This "pseudo-model" is refined and the
> R-factor and R-free are around 19%. Initial MIRAS phases are reliable and
> initial map is practically continuous in all its extension. My question is:
> Is there any program that uses this pseudo-model and the initial MIRAS map
> (or an omit-map) to validate the side chain and/or to guess the most probably
> side-chain? I am not sure if at 1.9 A I will be able to distinguish between
> Leu, Asn and Asp (as an example) solely by the electron density... Cysteine,
> Methione residues can be identified by the anomalous signal and the peak in
> the 2Fo-Fc map. Other residues are easily identified (Proline, Arginine,]
> Tyrosine ...) but what can I do distinguish between Leu/Asn/Asp, Gln/Glu,
> Val/Thr...?
> Amino acid sequence analysis is under way, but it may take a while since this
> is a big protein.
> Any help is welcome!
> All the best
> Ronaldo Nagem.

ANSWERS:

FROM: Anastassis Perrakis
I have a side_trace version that will guess the type. side_trace always makes
the guess actually, just does not print. But, better do manually!
Theres a paper by Hilge,Perrakis et al at ActaD describign some ideas for how to
do that. "Hilge, M., Perrakis, A., Abrahams, J.P., Winterhalter, K., Piontek, K.
and Gloor, S.M. (2001). Structure elucidation of B-mannanase: from the
electron-density map to the DNA sequence. Acta Cryst. D57, 37-43.

FROM: Marjorie Wilke
Take a look at Jane Richardson's Molprobity.  That might help.

FROM: Bart Hazes
At 1.9A you should be able to distinguish Leu from Asp/Asn. However, Asp/Asn,
Glu/Gln, Thr/Val will be difficult. In addition, you may have assigned the odd
alanine to a residue with a disordered sidechain. You can use hydrogen bonding
patterns to make some of the distinctions though.
I also suggest, if you didn't already do so, to throw your pseudo sequence
through a BLAST search. If there are some close homologs then you can use that
tentatively as "prior information" until you determine the real sequence
experimentally.

FROM: David Margulies
You might want to look at
Anderson CM, Stenkamp RE, Steitz TA. 1978. Sequencing a protein by x-ray
crystallography. II. Refinement of yeast hexokinase B co-ordinates and
sequence at 2.1 A resolution. J Mol Biol 123: 15-33
Anderson CM, McDonald RC, Steitz TA. 1978. Sequencing a protein by x-ray
crystallography. I. Interpretation of yeast hexokinase B at 2.5 A resolution
by model building. J Mol Biol 123: 1-13

FROM: H.C.A.Raaijmakers
Leu/(Asn/Asp) can be identified by the water structure, and the same for
val/thr. look for hydrogen bonds  shorter than 3.2 A. see whether it packs
to hydrophobic or hydrophilic parts and make a guess. it should be possible
to guess ~ 90% correct at this resolution. I once had 85% correct, 5%
indistinguishable side chains, (Asn/Asp etc) 5% disordered side chains and 5%
plain wrong, which could have been resolved with help of the water structure.
The sequence you find can help to design primers for sequencing. For the
refinement (R factor) it doesn't matter much if you have an asp or leu instead
of asn. Very handy is a blast search (sequence alignment), to see what similar
proteins look like, if they all have a val at a certain position it is unlikely
you have a thr. Use everything to get clearer maps, such as a good solvent
model,
add waters (~ 1 per amino acid at this resolution) , try refmac with TLS
refinement, try to use experimental phases (perhaps blurred or weighted down
in case the FOM is overestimated)
Of course it is a good option to work three months on something else, until
the sequence comes in. Instead of building three months now, you'll do it in
one or two weeks with a sequence. But don't forget to inquire once a week about
the progress of the sequence :-)