You are here

ligand docking, how to combine some silent.out files into a silent_all.out file

7 posts / 0 new
Last post
ligand docking, how to combine some silent.out files into a silent_all.out file

Hi, everyone!
I run ligand docking many times to obtain several silent.out files about same protein-ligand case. I want to combine several silent.out files into a silent.out to convenience next step analysis. I try my best to use "extract_atomtree_diffs" to generate pdbs about several silent.out files, and then combine pdbs into a new silent.out include all pdbs. It is fail because it is not supportted by next analysis. And I try to use "combine_silent" to combine them but again it is not success. So could you tell me how I can do it? I would be glad if you could help me with these, Thank you in Advance!

Post Situation: 
Fri, 2013-12-06 01:53
Ryhon Wang

The "simple" way of doing it is just to try concatenating the atom tree diff files together. (e.g. "cat silent1.out silent2.out silent3.out > silent_all.out" on the commandline.) If I recall correctly, this should work for atom tree diff files.

The other option is to convert the atom tree diff format files into "regular" binary silent files individually, and then combine the binary silent files. If you're running Rosetta 3.3 or later all of the (compiled) ligand docking related applications should be able to take binary silent files as input. To do the conversion (again, assuming Rosetta 3.3 or later) you can simply run the score_jd2 application with "-in:file:atom_tree_diff {inname}" and "-out:file:silent {outname} -out:file:silent_struct_type binary" flags.

The atom tree diff format, while compact, has deficiencies as a representation, and their use is deprecated in favor of binary silent files or collections of PDBs.

Fri, 2013-12-06 08:58

Thanks a lot, rmoretti!!!
I am glad to see you reply!

Thu, 2014-01-16 00:12
Ryhon Wang


I am not sure if this is the correct topic or not for my problem, but I did not want to start a new topic for one that already existed.


Anyway, I am having trouble combining silent files. I have 1000 silent files, each containing 25 decoys.

So this is the problem:

when I extract a PDB from one of the silent files the structure is perfect, no problems.

When I use the combine_silent.default.linuxgccrelease executable to combine all the silent files into one large silent file then extract the same structure it becomes corrupted.

Is this a bug? does it have a fix? I need all the decoys in one silent file so i can comfortably search through them.


I tried concatinating all silent files into 1 file, which works, and the extracted PDB is fine, BUT i get file_0001 repeated 1000 times and rosetta will only extract the first etiration of the name. Therefore I need all my decoys with unique names within the silent file, (does it make sense)?


Please tell me how i can fix this


Thank you

Mon, 2017-01-09 05:21

What do you mean by "corrupted"? The combine_silent approach should work.

The issue with concatenation is, as you mention, that the structure numbers repeat. I believe (but have not tested) that the structure renumbering that occurs with combine_silent also will occur with read-in, so you should be able to use the (e.g.) "file_0001_2" multiple-name rename name to get the subsequent files. Another alternative is simply to do something like a silent to silent minimal protocol to force the renumbering (e.g. something like score_jd2 with -out:file:silent) - though that should really be what combine_silent is doing, so I'm not sure how well that would work.

The other alternative, when you do re-runs, is to use the -out:prefix or -out:suffix options to use different labels on each run, such that the structures have distinct names on the output.

Mon, 2017-01-09 09:52

Hi rmoretti,


Thank you for your replay.

Yes it is strange, because i have used the combine_silent.default.linuxgccrelease executable several times, and this is the first time this issue occures.

I have 1000 silent files, they are in binary format, I think that when i combine them  all into one large silent file they get converted from binary format into a normal silent file format. I am guessing something is happening during this convertion that is causeing extracted PDB file to look bad (by that i mean you find imposibble configurations such as a strand going through a helix).


Is there a way to combine binary silent files and keep them in binary format?

Mon, 2017-01-09 12:54

I think that Rosetta should automatically recognize that the particular structure can't be represented with the protein silent file format and automatically choose the binary format, but if you want to force the issue it should be possible to add `-out:file:silent_struct_type binary` to the combine_silent commandline to force binary format on output.

Wed, 2017-01-11 12:53