[phenixbb] autobuilding homomers

Ingemar André Ingemar.Andre at biochemistry.lu.se
Mon Oct 29 11:18:48 PDT 2012

Hi Tom, 

Thanks for your input. I left out a few details: The monomers are structurally identical by construction. They adopt a perfect C3 trimer, so the NCS is spot on. If I run with rebuild_in_place=True on the best working model after running with rebuild_in_place=False (using the command suggested by phenix for "making a very good model") then the model still has asymmetric sequence. The sequence I give as input file is then not used during building. This probably due to differences between my given correct sequence in the sequence file and the sequence in the model generated by phenix. I get warnings like this:

NOTE: chain A (residues 71-76) AELVEL
did not match the input sequence...
ignoring alignment and rebuilding segment without sequence information.

If I follow your suggestion and take the best map from the best working model and run autobuild with it (rebuild_in_place=False) without any input model I get back a model with same type of issues that I described in my email (wrong sequence, chainbreaks etc). However, now there are more residues in chain B rather than A.
If I use rebuild_in_place=True directly on my starting trimer model the resulting phenix model has the correct sequence. However, the problem with rebuild_in_place=True is that the radius of convergence is quite small. So it would be great if was possible to use rebuild_in_place=False and directly get structures with the right sequence. A way around this problem can potentially be to take the only helix that is built with the correct sequence by phenix and apply the symmetry operations to generate the two other helices. Then I could run with rebuild_in_place=True on this new trimer. I have not tried this yet. There are other similar system we study so it would be great if a general approach can be applied to these type of systems.

Thanks for your help,


On Oct 28, 2012, at 3:40 PM, Terwilliger, Thomas C wrote:

> Hi Ingemar,
> This is an interesting case.  It is a little difficult to tell exactly why your model does not end up with 3 more-or-less identical monomers.  The basic question will be: "Is there really near-perfect NCS in the structure?".  Here are some things to try:
> 1. Take your best model and run phenix.find_ncs with that model to get an NCS parameters file (find_ncs.ncs_spec).  Check that this file does have 3 NCS copies.
> 2. Try applying the ncs from find_ncs.ncs_spec to your "best" monomer to generate a trimer:
>  phenix.apply_ncs my_best_monomer.pdb find_ncs.ncs_spec
> Now look at this "perfect" trimer.  Do the 2 generated monomers actually match up with parts of the original model?  Or instead, is the NCS not really so perfect?
> 3. Try running autobuild starting from your "perfect" trimer, with rebuild_in_place=False (or also try with rebuild_in_place=True).
> 4. Take your best map from your best working model, throw away the model, and run autobuild starting from the map with input_map_file=my_best_map.mtz and let it build everything from scratch.
> All the best,
> Tom T
> ________________________________________
> From: phenixbb-bounces at phenix-online.org [phenixbb-bounces at phenix-online.org] on behalf of Ingemar Andre [ingemar.andre at biochemistry.lu.se]
> Sent: Sunday, October 28, 2012 6:35 AM
> To: phenixbb at phenix-online.org
> Subject: [phenixbb] autobuilding homomers
> I have a question about the proper procedure for building homomers. My
> test case is a trimeric coiled-coil with 3-fold symmetry (trimer in
> assymetric unit). My input model is a trimer with three chains (A,B,C)
> with identical sequence. If I run phenix autobuild
> (rebuild_in_place=False) with this model I get a structure with decent
> R/Rfree. However, there are some issues:
> 1) Helix 1 consists of chain A in the output structure. There is a
> jump in the sequence number for chain A at the end. However,
> disregarding this chain break the full sequence is correct for helix 1.
> 2) Helix 2 consists of chain B + C in the output structure. The chain
> B has the correct sequence for helix 2. Chain C has partly the right
> sequence for helix 2.
> 3) Helix 3 consists of chain C in the output structure. Chain C has
> partly the correct sequence for helix 3. Lots jumps in the sequence
> numbering.
> GIven that I know its has three subunits with identical sequences I
> gather it should be possible to have autobuild generate structures
> with three chains with identical sequences. I have tested a few
> things. With and without giving it the sequence file corresponding to
> the sequence of a monomer. With and without enforcing ncs based on the
> input model (find_ncs=False input_ncs_file=find_ncs.ncs_spec). I am
> never able to build sequence-symmetric structures with
> rebuild_in_place=False for any homomer I have tried to build. I have
> limited experience with phenix so it is likely I am missing something
> obvious. In any case, I would appreciate your input on this.
> Best regards,
> Ingemar
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb

More information about the phenixbb mailing list