Hi Tom, thank you very much for your valuable suggestions. best wishes Alejandro On 2/23/14, 21:59, Terwilliger, Thomas C wrote:
Hi Alejandro,
I'm glad that phenix.fit_loops is now doing what it is supposed to (and I did fix the documentation now too.)
On your question of how to handle the loops with poor density: there isn't a good automated way to do this that I know of. What I would recommend: build the loops with phenix.fit_loops or some other tool like it. Remove all the ones that really do not fit the density at all (and try again with a different random seed, or more tries). For the remaining ones, try autobuild rebuild_in_place, specifying these loops for rebuilding, and see if you can get them improved automatically. You can do that a few times too if you want, also with varied random seeds. For those that don't improve, try doing it manually as you mention. If you can't fit them well, it is most likely because there really are many conformations. This is a situation that we don't yet have a good way to model, particularly at a resolution of 2. 75 A, and I would say that as a practical matter today you're pretty much stuck with what you can represent with a single model. I would recommend leaving out any loops that you cannot model in a satisfactory way, despite that fact that this leaves an incomplete model.
All the best, Tom T
________________________________________ From: Alejandro Buschiazzo [[email protected]] Sent: Sunday, February 23, 2014 11:11 AM To: Terwilliger, Thomas C; [email protected] Subject: Re: [phenixbb] revisiting "Problem with loop fitting "
Dear Tom,
it worked! thanks a lot
I followed your suggestions first with one loop (I have six monomers in the ASU, so I have say 12-18 loops I would like to build in): the program chose the best fit with CC of 0.54. ...and a couple of Ramachandran outliers appear now in this added loop as well.
So, I would like to ask for your advise here, if I may; just a very brief explanation of the specific problem:
this is a case where the density clearly says a loop should be built in, but simultaneously, is not at all a "good density" (this is why the loops are actually missing, at this, rather advanced, point of the refinement). Let me just add that resolution is 2.75A, but I have 0.42 hemihedral twinning in P65 (twin law k,h,-l)...best crystal and data we could get after lots of work: so, maps are good, but definitely not great.
If I attempt adding these loops manually, it is not easy...and typically end up having too many Ramachandran outliers (and don't want to add phi/psi restraints in refinement)
I would so much appreciate any hint, as to the best way to handle this kind of situation to automatically build at least stereochemically correct loops, in clear but not easily interpretable electron density.
Many thanks again Tom for your help
Alejandro
On 2/23/14, 12:39, Terwilliger, Thomas C wrote:
Hi Alejandro,
I think that the documentation is not clear enough about how to specify the gap (I will fix this). Try this test:
phenix_regression.wizards.test_command_line_loops test_fit_loops
Then modify the command in that test:
phenix.fit_loops pdb_in=nsf_gap.pdb mtz_in=resolve_0.mtz seq_file=nsf.seq start=37 end=43 chain_id=None
to match your PDB file and mtz file.
The confusing thing (I think) is that start=37 means the first residue in the gap is 37, not the last residue already present is 37.
Let me know if that doesn't do it! All the best, Tom T
________________________________________ From: [email protected] [[email protected]] on behalf of Alejandro Buschiazzo [[email protected]] Sent: Saturday, February 22, 2014 2:28 PM To: [email protected] Subject: [phenixbb] revisiting "Problem with loop fitting "
Dear phenix community
I am attempting to use phenix.fit_loops for the first time and getting an error message: "No suitable gap found matching residues"
I found a short exchange exactly about this (Jan 2011), but eventually Tom Terwilliger asked for the files directly to see if he could solve it. I can't seem to find a later summary of the solution.
(apparently, the most probable reason was some kind of mismatch between PDB numbering and sequence file...which I shouldn't have... I'm probably missing some silly here - I have been fiddling around with the sequence file in many ways...with no success)
many thanks for any help! alejandro
-- Alejandro Buschiazzo, PhD Research Scientist Unit of Protein Crystallography Institut Pasteur de Montevideo Mataojo 2020 Montevideo 11400 URUGUAY Phone: +598 25220910 int. 120 Fax: +598 25224185 http://www.pasteur.edu.uy/pxf
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
-- Alejandro Buschiazzo, PhD Research Scientist Unit of Protein Crystallography Institut Pasteur de Montevideo Mataojo 2020 Montevideo 11400 URUGUAY Phone: +598 25220910 int. 120 Fax: +598 25224185 http://www.pasteur.edu.uy/pxf
-- Alejandro Buschiazzo, PhD Research Scientist Unit of Protein Crystallography Institut Pasteur de Montevideo Mataojo 2020 Montevideo 11400 URUGUAY Phone: +598 25220910 int. 120 Fax: +598 25224185 http://www.pasteur.edu.uy/pxf