Dear phenixbb,

I would like to use mr_rosetta to build a second protein chain.

For the first protein, I have a search model with known structure. The search model can be placed with Phaser (TFZ 10). For the second protein, the structure of a homologue has been determined. The second protein accounts for about 2/3 of the residues in the ASU.

I would like to fix the first protein chain and build the second protein chain with mr_rosetta. I have obtained fragment files for both proteins from robetta.

What input parameters are required to tell mr_rosetta to build the second chain only? I have tried the fixed_ensemble_ID_list = ensemble1 option.

Here is an excerpt of my parameter file with the input files:

run a)

search_models = "path/chainA.pdb" (from Phaser)

search_models = "path/chainB.pdb" (homologue structure)

fixed_ensembles {
�� fixed_ensembleID_list = ensemble1
�� fixed_euler_list = 0.0 0.0 0.0
�� fixed_frac_list = 0.0 0.0 0.0
�� fixed_frac_list_is_fractional = True
�� }

alignment_files = "path/chainA.dat"
alignment_files = "path/chainB.dat"

with chainA.dat having twice the same sequence and chainB.dat with the target and template alignment.

I have set fragment files with chain_list for both chains.

I get this error when running phenix.mr_rosetta (running from command line with .eff file)

"Sorry, the file /path/GROUP_OF_PLACE_MODEL_1/RUN_2/chainB_al_0.pdb representing group 0 has no atoms?
This could indicate a problem with sequence alignment in rebuilding.
It could be that there are near-duplicate sequences in your sequence file
...if this is the case then you may need to artificially rename your duplicated
chains and renumber residues to make a single chain containing a duplication"

This error also occurs
�

run b) Without any alignment files

run c) Without any alignment files and with just chainA.pdb as search model.

An alignment of the sequences from chainA and chainB gives low identity (no duplicate regions), so I suspect the problem is located somewhere else?

I have seen a post by Tom Terwilliger, explaining how to deal with two identical chains with different conformations where no alignment files are required. However, I would like to use the information from the homologue structure, that is why I specified the alignment file.

By inputting chainB.pdb as search model, mr_rosetta also tries to use this model as a search model for molecular replacement, but I think this is not necessary (and gives a warning that the space group of the search model and the data file do not match). I just want to provide the structural information from structure chainB.pdb to mr_rosetta for building.

-----

So in summary this is what I would like to do:

chain A: keep fixed -� use chainA.pdb as fixed model
chain B: build - use information from homologue structure (chainB.pdb) and fragment files to build chain

What is the best strategy to do this? What input parameters are required so that I do not get the error described above?

-----

Thanks a lot in advance,

Tobias.