You are here

How does one prevent Rosetta from connecting two separate chains?

8 posts / 0 new
Last post
How does one prevent Rosetta from connecting two separate chains?

I have a homodimer of two separate chains, each one 31 residues long. I am attempting to run a comparative modeling protocol. Every pdb output has the two chains connected, which screws up subsequent relaxation.

How do I change the input fasta file, align file, and/or flags file to indicate that there are two separate chains?


Current files:
fasta file:
> Tm7_37

alignment file:

command file:
/sware/rosetta/linux/rosetta_source/bin/minirosetta.linuxgccrelease @Tm7_37.flags -database /sware/rosetta/linux/rosetta_database -run:constant_seed -nodelay -out:overwrite true > Tm7_37RosCM.log

flags file:
-run:protocol cm

-in:file:fasta Inputs/Tm7_37.fasta
-in:file:alignment Inputs/Tm7_37.2zta.aln
-in:file:template_pdb Inputs/2ZTAnoHET.pdb
-in:file:extended_pose 1
-out:nstruct 1

-loops:frag_sizes 9 3 1
-loops:frag_files Inputs/aat000_09_05.200_v1_3 Inputs/aat000_03_05.200_v1_3 none
-loops:remodel quick_ccd
-loops:relax fastrelax

-cm:min_loop_size 4
-cm:loop_close_level 0

-out:file:silent_struct_type pdb

-unmute protocols:looprelax




Post Situation: 
Thu, 2011-12-15 14:08

By the way, adding "-read_all_chain" to either the command file or flag file reulted in an ERROR

ERROR: Option matching -read_all_chains not found in command line top-level context

Fri, 2011-12-16 10:03

Is it expected to be a symmetric homodimer? Can you predict on one chain and dock? What about fold_and_dock? I don't think the basic ab initio app can do this.

Sun, 2011-12-18 16:08

It is expected be (part of ) a parallel alpha-helical homodimer, being nearly in-register but having a small (~1 Angstrom) axial offset from perfect two-fold symmetry.The two alpha-helical chains are intimately interacting with one another throughout their lengths in a coiled coil.

I never even attempted ab initio modelling for a number of reasons, foremost is that there is not enough information in the sequence of the peptide to indicate that it is a near in-register parallel coiled coil. (In vivo, this peptide is a fragment of a much longer sequence, and the information for the near in-register and parallel nature of the coiled coil resides elsewhere, beyond the fragment.)

Hence, I had thought to do a comparative modeling, based on a known structure that is a parallel in-register homodimer, and then, ?in the cm protocol? relax/refine the structure to see what Rosetta predicts, i.e. would Rosetta introduce the expected 1.0 Angstrom axial shift.

It seems a bit unusual that I can't figure out how to tell Rosetta's comparative modelling that the two chains are not covalently attached to each other, but in any case, perhaps I could try docking one chain to the other. Do you think that docking one alpha-helix to another to form an initmate coiled coil is appropriate for docking? Also, do you have a Rosetta3.3 compatible docking protocol that takes as input a modeled structure that is close to correct?

Thanks so much, -Jerry

Tue, 2011-12-20 07:59

You're right that for a tightly coiled-coil, docking is unlikely to work well.

Ab initio may well work fine on this problem - the issue is that the basic abinitio_relax executable isn't used much by the developers anymore, so nobody knows its quirks, and the newer, fancier abinitio protocols are either unreleased or undocumented.

I've sent this along to one of the comparative modeling people who may know a way to do it...

Wed, 2011-12-21 17:09

The answer seems to be, "you can't get there from here". Ab initio is meant for single-domain single-chain predictions. I guess you can try something like adding a faux GSGSGS linker and then just deleting it afterwards...?

Wed, 2011-12-21 17:58

Thanks. I had thought of putting in a long fake linker, but was hoping to avoid it. But perhaps it is worth a try.

Thu, 2011-12-22 12:47

Thanks smlewis, adding a faux GSGSGS linker and then just deleting it afterwards has worked out! I was trying to follow the initial guidelines provided by the topic starter but failed. I was in a hurry to fix the issue so my work really resembled typing speed test :) Now it works well, thanks a bunch!

Fri, 2012-06-01 00:40