# Clustering using an disrupted run silent file

10 posts / 0 new
Clustering using an disrupted run silent file
#1

Hi all,

I was running rosetta but something happened and the job was stopped. I would like to use the already generated silent file to cluster but when I run the clustering algorithm I get a 'could not make' error which I suppose is due to the incomplete status. Had set to generate 10,000 but at the point of interruption there were 4767 structures. How do I edit the silent file for proper clustering of the 4767 decoys?

Post Situation:
Mon, 2013-07-15 02:19
PaulaBanks

Hi Paula,
Try after deleting the lines for the decoys for which the run was incomplete, if it actually appears in the silent file. If possible please provide the command you used and the error message you got.

Mon, 2013-07-15 04:22

Command: ../rosetta_source/bin/cluster.linuxgccrelease @flagsc > cluster.log

My flagsc:
-database ../rosetta_database
-in:file:silent singh_silent.out
-in::file:s
-in::file:fullatom
-cluster:sort_groups_by_energy
-out:file:silent

ERROR: Could not make
ERROR:: Exit from: src/core/io/silent/SilentFileData.cc line: 287

I also looked at the silent file but I see that the last run had gone through my 71residues like in the 4766 run.

Wed, 2013-07-17 01:34
PaulaBanks

-out:file:silent takes a filename. The error you're seeing is Rosetta complaining that it can't create a file with the empty string as a name.
Use something like "-out:file:silent silentfile.out" instead.

Wed, 2013-07-17 11:13
rmoretti

The command line flag "-silent_read_through_errors" may also help to recover a corrupted silent file. (Just add it to any Rosetta run using the silent file as input.)

Mon, 2013-07-15 11:56
rmoretti

Hi Moretti,

I added the flag '-silent_read_through_errors' but I still get the same error.

ERROR: Could not make
ERROR:: Exit from: src/core/io/silent/SilentFileData.cc line: 287

Wed, 2013-07-17 01:31
PaulaBanks

Hi Paula,
Not sure whether this will work,but please run it after deleting 'in:file:s'
flags:

-database ../rosetta_database
-in:file:silent singh_silent.out
-in:file:fullatom
-cluster:sort_groups_by_energy
-out:file:silent

Rocco, could you please explain the difference between in:file:fullatom and in::file:fullatom ?

Thanks.

Wed, 2013-07-17 02:57

There's none. The option parsing machinery should treat "::" and ":" identically. Which to use is a matter of preference. (The double colon matches the C++ namespace resolution of the options, but there was historically some issues (long since fixed) with Mac (?) computers where ':' would work and '::' wouldn't.)

Wed, 2013-07-17 11:09
rmoretti

Hi,
I also try to work with the cluster tool in the same way like Paula. But the problem is, that Rosetta doesnt find the structures and clusters. I have define the structures with the commando : -in:file:native .input/F_000*.pdb \. Whats wrong?!

Wed, 2014-09-24 06:09
masterofpuppets

What's the full command you're using, and what's the error you're getting? (Copy and paste tends to make debugging easier than paraphrasing.)

One possibility is that -in:file:native is expecting a single structure which is the native file to which RMS statistics are calculated - other than being used to calculate RMSD, the native file shouldn't affect the clustering.

If you want to cluster PDBs, you need to use the "-s" option to specify them. So if you wanted to cluster the structure named like input/F_000*.pdb, then you'd use the option "-s input/F_000*.pdb" instead of Paula's "-in:file:silent singh_silent.out" (which she used because she had a silent file instead of a PDB as input.)

Note that you need to give the names of the PDBs that you want to use - just a directory name won't work. Also, remember it's the shell that's doing the wildcard ("*") expansion, not Rosetta, so if you can't do a "ls .input/F_000*.pdb" and get a list of filenames, then Rosetta won't be able to find them.

Wed, 2014-09-24 09:57
rmoretti