You are here

I'm a total beginner to Rosetta 3.2 and there are no tutorials for me to follow. Can someone help.? =)

13 posts / 0 new
Last post
I'm a total beginner to Rosetta 3.2 and there are no tutorials for me to follow. Can someone help.? =)
#1

Hi, I'm currently an undergraduate research student in Singapore. My prof has just tasked me to try out Rosetta for docking. However, he is also new to the program, and so we are sort of stuck on how to start even a basic docking. We are using SUSE Linux. We hope that someone would be kind enough to get us started on a very basic tutorial on docking like the commands used, before we proceed on to explore the features further.

Thanks. =)

Post Situation: 
Wed, 2011-02-02 08:11
monos_morpheus

Did you get it to compile?

Assuming you did, start here:
http://www.rosettacommons.org/manuals/archive/rosetta3.2_user_guide/Dock...

There may be demos included with the release; the integration tests (test/integration/tests/(anything that starts with docking_)) also serve as demos.

Anything that's not in that guide...you can ask further questions here, or email the corresponding author (Jeff Gray) directly. They rebooted the docking executeable for 3.2, so he needs to know if they've still got documentation flaws so they can get fixed...

Wed, 2011-02-02 08:51
smlewis

Hi, thanks, by compiling, I presume you would be referring to the .linuxgccrelease files? If yes, I did. However, in the webpage you provided, [path to executable]/docking_protocol.[platform|linux/mac][compile|gcc/ixx]release –database [path to database] @options, how do i input the pdb files to be docked? Is it okie if you give an example for this command, for the case where R.pdb is the receptor and L.pdb is the ligand? Many many thanks.! =)

Wed, 2011-02-02 20:08
monos_morpheus

This will copy one of the docking demos and run it for you. I can't give you the correct paths, you'll have to fix those.

cp -r (whatever)/rosetta_source/test/integration/tests/docking_full_protocol wherever/you/want/to/put/it
cd wherever/you/want/to/put/it
cat flags (then read the flags to see your answer for a command line)
path/to/rosetta_source/bin/docking_protocol.linuxgccrelease @options -database path/to/rosetta_database/

The @flags syntax allows you do put all the command line flags into a file, instead of cramming them onto the command line itself. The PDB input is passed in via the "-s" flag. You'll note the input already contains both receptor and ligand. For this trivial input, they start out already docked, but you can just concatenate your two PDBs and that will work fine as a starting structure. (The spin flag tells Rosetta to randomize the orientations of the input chains, so Rosetta is not "cheating" by starting at the correct answer in this demo).

Thu, 2011-02-03 07:45
smlewis

Oh ok.! Thanks.! Sorry, I have 1 more qn, how do i process the silent.out file ater doing ligand docking.? =)

Thu, 2011-02-03 08:05
monos_morpheus

The easy way is to figure out the flag that has it dump PDB output instead of regular output. I don't know the flag, check the manual and let me know if you don't find it. It may default to PDB so you may have a flag already there that redirects to silent output.

Failing that, try http://www.rosettacommons.org/manuals/archive/rosetta3.2_user_guide/sile... extract_pdbs.

I KNOW that silent files are better than PDBs for most rosetta applications, but I still refuse to use them, because they're such a hassle...

Thu, 2011-02-03 17:17
smlewis

haha, i totally agree on that.!

This is one of the options in the flag for ligand docking:

-out
## These channels generate a LOT of output,
## especially with -improve_orientation.
-mute protocols.geometry.RB_geometry core.scoring.dunbrack.SingleLigandRotamerLibrary core.scoring.rms_util
## Number of trajectories to run (per input structure given with -s)
-nstruct 500
## With multiple processes, ensures a unique name for every output structure.
## Each process should get a different string here, so you can't really
## put it in the FLAGS.txt file, it has to go in the Condor script:
#-suffix 3
-file
## I prefer output structures with Rosetta numbering, from 1 to N residues.
## To keep the original PDB numbering, omit this flag:
-renumber_pdb
-path
## Where to write the silent.out file. Specified in my Condor script.
#-pdb work/jnk_pp_1_001/3

So, I don't think I can change it to output as pdb files instead. And using the extract_pdbs.linuxgccrelease, it says "bad format in first line of silent file", then the process stopped after a while, not generating any pdb files. What can I do to resolve any of these issues? =D

In addition, is there any difference in using extract_pdbs or extract_atomtree_diffs.linuxgccrelease? Also, I encountered a 'unrecognized aa 100' error when using ligand_rpkmin.linuxgccrelease, what does it mean?

Fri, 2011-02-04 07:03
monos_morpheus

"## Where to write the silent.out file. Specified in my Condor script.
#-pdb work/jnk_pp_1_001/3"

Um...that looks like it's putting out PDBs already. Are you sure they're silent files? Post one as an attachment.

"Also, I encountered a 'unrecognized aa 100' error when using ligand_rpkmin.linuxgccrelease, what does it mean?"
You have an input, probably a PDB, misformatted such that it has a residue named 100 (instead of numbered 100). The PDB file format is column-based, so which column the characters appear in determines how they're interpreted. The PDB website has the official specifications someplace.

Fri, 2011-02-04 07:16
smlewis

Okie, because the forum doesn't allow me to post .out file, I have added a .jpg extension to it instead, just download the file and rename it to remove the .jpg extension. Oh okie, I will check on the pdb file again. You must have been using Rosetta for a long while before knowing all these.!

Yah, in my ligand file, the residue is '100', however, what can I changed it to for the respective atoms since the ligand is not a residue?

Fri, 2011-02-04 08:04
monos_morpheus

For the silent file:

It looks like this silent file format (there are at least 5, none of which are properly documented) contains regular PDBs! Try just grepping (you know, command line grep) the ATOM and HETATM lines.

For your ligand:
Do you mean it is named 100 in the parameters file, or in the a PDB file? Either way, you can just change the residue name to whatever three-character string you want. I don't really understand the phrasing of your question.

Fri, 2011-02-04 13:02
smlewis

Overview and checklist
Here is a "checklist" for setting up a ligand docking experiment. Details for all the steps are given below.
A. Reformat starting structures
1. Protein receptor in PDB format with no waters or cofactors; each metal ion should have a unique chain ID.
a. Does this mean that from the PDB file downloaded, we just need to delete the waters and metal ions, and just keep the protein residues instead?
2. Cofactors, if any, should be in .sdf or .mol2 format, with coordinates taken from the original PDB file.
a. Does this mean that I should make a copy of the original PDB file, then delete all the lines just leaving the cofactor (metal ion), and then convert to sdf or mol2 using Babel?
3. Ligands to be docked should be in .sdf or .mol2 format, one ligand per file. If you're trying to reproduce a known binding mode, those ligand coordinates should be used.
a. Does this mean that I should make a copy of the original PDB file, then delete all the lines just leaving the ligand, and then convert to sdf or mol2 using Babel?
B. Prepare ligands and cofactors
1. Generate conformers if desired. All conformers should be in the same .sdf / .mol2; one file per compound. Tautomers / protomers count as different compounds.
2. Assign partial charges if desired. Only charges from the first entry in the file are used. Charges can only be read from .mol2 files.
a. If I do not have Openeye, does that mean I --skip-omega and --skip-charges?
3. Convert the .sdf / .mol2 into a .params file and one or more .pdb files. The .params describes atom types, charges, and rotatable bonds; the .pdb files contain ligand coordinates.
4. Make a ligand conformer library if desired. Concatenate the desired coordinates in PDB format and add a "PDB_ROTAMERS ..." line to the .params file.
C. Prepare the receptors
1. Repack the entire receptor in the presence of cofactors but the absence of ligands.
a. Here, when we use ligand_rpkmin.linuxgccrelease, do we use it on the ligand PDB or the protein PDB?
2. Choose the lowest-energy result to use as the repacked structure.
3. I repack by running "relax" with the sequence option.
D. Generate input files for Rosetta
1. Ligand .params files
2. Ligand conformer libraries (typically *_confs.pdb)
3. Input PDB: repacked protein + generated ligand conformation
4. Native PDB: experimental protein and ligand coordinates (optional, for RMSD calculations)
5. Unbound PDB: experimental protein without ligand (optional)
6. Rosetta flags file (typically FLAGS.txt)
E. Run docking trajectories
F. Analyze results
1. Concatenate silent.out files to get one per ligand/receptor pairing.
2. Extract score and RMSD information; make scatterplots.
3. Cluster ligand conformations.
4. Estimate docking confidence.

Fri, 2011-02-04 20:04
monos_morpheus

Overview and checklist
Here is a "checklist" for setting up a ligand docking experiment. Details for all the steps are given below.
A. Reformat starting structures
1. Protein receptor in PDB format with no waters or cofactors; each metal ion should have a unique chain ID.
a. Does this mean that from the PDB file downloaded, we just need to delete the waters and metal ions, and just keep the protein residues instead?

Delete everything that's not a protein atom. Rosetta ignores all the stuff at the beginning, may as well delete it too...

2. Cofactors, if any, should be in .sdf or .mol2 format, with coordinates taken from the original PDB file.
a. Does this mean that I should make a copy of the original PDB file, then delete all the lines just leaving the cofactor (metal ion), and then convert to sdf or mol2 using Babel?

I guess so. PyMOL can do it too.

3. Ligands to be docked should be in .sdf or .mol2 format, one ligand per file. If you're trying to reproduce a known binding mode, those ligand coordinates should be used.
a. Does this mean that I should make a copy of the original PDB file, then delete all the lines just leaving the ligand, and then convert to sdf or mol2 using Babel?
B. Prepare ligands and cofactors
1. Generate conformers if desired. All conformers should be in the same .sdf / .mol2; one file per compound. Tautomers / protomers count as different compounds.
2. Assign partial charges if desired. Only charges from the first entry in the file are used. Charges can only be read from .mol2 files.
a. If I do not have Openeye, does that mean I --skip-omega and --skip-charges?

I have no idea what you're asking, sorry...

3. Convert the .sdf / .mol2 into a .params file and one or more .pdb files. The .params describes atom types, charges, and rotatable bonds; the .pdb files contain ligand coordinates.
4. Make a ligand conformer library if desired. Concatenate the desired coordinates in PDB format and add a "PDB_ROTAMERS ..." line to the .params file.
C. Prepare the receptors
1. Repack the entire receptor in the presence of cofactors but the absence of ligands.
a. Here, when we use ligand_rpkmin.linuxgccrelease, do we use it on the ligand PDB or the protein PDB?

protein

2. Choose the lowest-energy result to use as the repacked structure.
3. I repack by running "relax" with the sequence option.
D. Generate input files for Rosetta
1. Ligand .params files
2. Ligand conformer libraries (typically *_confs.pdb)
3. Input PDB: repacked protein + generated ligand conformation
4. Native PDB: experimental protein and ligand coordinates (optional, for RMSD calculations)
5. Unbound PDB: experimental protein without ligand (optional)
6. Rosetta flags file (typically FLAGS.txt)
E. Run docking trajectories
F. Analyze results
1. Concatenate silent.out files to get one per ligand/receptor pairing.
2. Extract score and RMSD information; make scatterplots.
3. Cluster ligand conformations.
4. Estimate docking confidence.

Mon, 2011-02-07 12:54
smlewis

It's okie, I managed to figure it out by myself and get it running, nonetheless u r a great help and I foresee I will need more help from in the very near future, lol. =)

Tue, 2011-02-08 09:31
monos_morpheus