You are here

No output pdb files after docking protein and DNA/RNA

12 posts / 0 new
Last post
No output pdb files after docking protein and DNA/RNA
#1

I tried to dock the protein and DNA or RNA with perturbation flags, everything seems going well, but at the end of the docking, these is no output pdb files. Below is the information at the end of the log file, and also the flags are shown.

protocols.docking.DockingProtocol: Scoretype: interchain_contact score: 12, cutoff: 9.99
protocols.docking.DockingProtocol: Scoretype: interchain_vdw score: 0, cutoff: 1
protocols.jd2.FileSystemJobDistributor: Too many retries (max_retry_job = 10)
protocols.jd2.JobDistributor: 10MH_0001 reported failure and will retry
protocols.jd2.JobDistributor: no more batches to process...
protocols.jd2.JobDistributor: 1 jobs considered, 11 jobs attempted in 66 seconds

-database /home/zyan/rosetta3.4/rosetta_database/
-s /vault/zyan/SPA-PN/NativePdb/10MH.pdb
-partners A_BC
-nstruct 1
-ex1
-ex2aro
-dock_pert 3 8
-use_input_sc
-overwrite

If I change the perturbation flag "-dock_pert 3 8" to refinement flag "-docking_local_refine", it works well and the pdb file is produced. How to deal with this problem? your response and help is very appreciated.

AttachmentSize
log.txt1.92 MB
Post Situation: 
Thu, 2012-10-25 08:23
yzq

It's failing the docking filters. This line: " protocols.jd2.FileSystemJobDistributor: Too many retries (max_retry_job = 10)" explains the problem.

You can try increasing the number of attempts with -ntrials. It defaults to 10. You can try as large a number as you like. I don't have enough information about _why_ docking's filters are failing to tell you if increasing ntrials is worth the trouble, or if we need to do something else to get the centroid-mode filters to pass. I've sent this along to the docking folks for comment.

Thu, 2012-10-25 11:12
smlewis

Looking in the code, it is also possible that the scorefunction term "interchain_vdw" used with this code fails to recognize the DNA/RNA - it is a centroid mode scorefunction term, and those usually don't work with DNA/RNA. I dug into its code and found:

if ( !rsd1.is_protein() || !rsd2.is_protein() ) return;

The short of this is: you can't do centroid mode docking with nucleic acids.

You can try doing fullatom docking (that's what happens when you use -local_refine_only), but the scorefunction is probably wrong for that too. I've contacted one of the DNA folks to ask how they do protein-DNA docking.

Thu, 2012-10-25 11:25
smlewis

Thank you, looking forward to your further help on this.

BTW, how to set up the flag ntrials for more retries? is it -jd2:ntrials 100? I tried this, but it seems not work.

Thu, 2012-10-25 12:28
yzq

It should be -jd2:ntrials. By "not work" do you mean it crashes with a message that the option ntrials is not found, or it continues failing to produce structures because the docking filters are failing (failing with similar messages as before)? Due to what I found in my second post, about the scorefunction, it is probably never going to work.

Thu, 2012-10-25 12:54
smlewis

It is still 10 tries not 100 tries. You mean RosettaDock can't sample the decoys of protein-DNA/RNA complexes by perturbation at current version3.4?

Thu, 2012-10-25 15:58
yzq

Is there a flag to disable the filtering for producing the output pdb files? I mean is there a way to skip the filter and directly output the pdb conformation files which are what I need.

In the log file, there is information below:

protocols.docking.DockingInitialPerturbation: moving together failed. Aborting DockingSlideIntoContact.

How to deal with this problem? I needn't the Rosetta energy score of the protein-DNA interactions, but the conformation generated by Rosettadock. Thank you

Fri, 2012-10-26 07:24
yzq

I don't think the standard docking executable can handle DNA or RNA as-is. The centroid phase can't score RNA/DNA. It doesn't matter that you don't want the scores - Rosetta can't make conformations at all without some ability to check that they are sensible.

You can try commenting out the filter that is failing - in particular, comment out (add // to the start of) lines 326-330 of src/protocols/docking/DockingInitialPerturbation.cc, and/or delete the portion " && pose.energies().total_energies()[ scoring::interchain_vdw ] < 0.1" of line 321. I'm not sure if this will actually work or not - it may have reasonable conformations it can't score properly, or it may have garbage conformations it can't score properly.

You can also just continue running the fullatom-only docking_local_refine mode that works.

Fri, 2012-10-26 07:41
smlewis

The Protein and DNA still don't move together after commenting out what you suggest. I don't know why.

Fri, 2012-10-26 15:49
yzq

I don't know either - that stage of RosettaDock was never meant to work with nucleic acids. In the short term, I guess use the fullatom-only docking, until I can get the attention of the DNA modeling folks.

Sat, 2012-10-27 09:19
smlewis

Hello everyone,

I also have this problem, I'm trying todock a protein with an ssDNA using the full protocol flags :

-s protein_ssDNA.pdb
-partners A_H
-database /home/useradmin/Rosetta/main/database/
-ex1
-ex2aro

 

But, I obtain the message error :

protocols.docking.DockingProtocol:  Scoretype: interchain_contact score: 12, cutoff: 9.99
protocols.docking.DockingProtocol:  Scoretype: interchain_vdw score: 0, cutoff: 1
protocols.jd2.FileSystemJobDistributor: Too many retries (max_retry_job = 10)
protocols.jd2.JobDistributor: HBsAg_Aptamer_0001_0001 reported failure and will retry
protocols.jd2.JobDistributor: no more batches to process...
protocols.jd2.JobDistributor: 1 jobs considered, 11 jobs attempted in 68 seconds


Can anyone know the solution to overpass this problem and realize the docking protocol protein/ssDNA ?

Mon, 2016-05-30 02:05
JulienO

The short answer is that you can't, at least with the current version of Rosetta.

The problem, as described above, is that the core evaluation metric that RosettaDock uses in the low-resolution phase just doesn't recognize nucleic acids. It ignores them completely. This means you can't use the low-resolution phase to dock proteins and nucleic acids - Rosetta just can't evaluate how goof the models are. 

The work around is to skip the low resolution stage, and go directly to the local refinement. The fullatom score functions used there should be able to handle the protein/nucleic acid interactions. The drawback is that you don't get the global sampling of the low resolution stage - you have to place the protein in approximately the binding conformation (to within a couple of Angstroms) before starting. 

There's currently efforts by Rhiju Das's lab at Stanford to improve Rosetta's ability to look at protein/nucleic acid interaction prediction, but as far as I know, that work is still in the "in progress" stage.

Fri, 2016-06-17 10:22
rmoretti