You are here

Calibur-linux binary or source code availability?

3 posts / 0 new
Last post
Calibur-linux binary or source code availability?
#1

Just wondering if there been any progress toward building the Cailibur program for clustering into RosettaDock?

Is the Calibur source code available anywhere?  All I seem to find is the binary version on the  http://sourceforge.net/projects/calibur/, and I only get this to work on a windows envrionment.  It would be nice to have Calibur version for use on  a Linux platform.

However, since we generate a Rosetta decoy set for protein-protein docking which involves a global (low-resolution-only) search, we need clustering that calculates rmsd's without superposition since one would just get all zero's, with superposition.  Rosetta's Cluster module allows this by using the "-cluster:skip_align" keyword.  Kind of not sure if calibur offers such option.  In particular, I notice that calibur (default) finds two large clusters compared to many smaller clusters using Rosetta Cluster with the same decoy set of ~5000 selected decoys.

 

Thanks in  Advance

Category: 
Post Situation: 
Sat, 2016-12-10 23:36
jasnyderjr

There has been progress - I think it will be available in the next weekly release.  (Note that releases are currently on hold as we prepare a major change to RosettaScripts).  

 

It won't be natively connected to docking and I don't know what support it has for various clustering styles.

Sun, 2016-12-11 07:11
smlewis

As Steven has said,

This will finally be in the next weekly release as an application and it will be compiled when you build Rosetta, available just like any other Rosetta app.  The application can certainly work with docking, but can not currently be called in Rosetta scripts.

If you need calibur immediately, that site that you linked DOES have a tar.gz that has the compilable files, makefile, etc.

https://sourceforge.net/projects/calibur/?source=navbar

                           Download calibur.tar.gz               

Calibur will print all of the clusters (at least the new version in Rosetta does this.  It will first give you the two largest, and then all of them.)  I'm not sure about the skip-align part.  I guess you don't want the superposition only because calibur is working on a single chain (otherwise, this shouldn't be a problem - so double check that you are giving calibur the correct chain)? I really don't know if calibur is doing a superposition ( I would go back and reread the primary citation, and I HIGHLY doubt that it is), but if it is, I can look into an option to turn off any superposition.  Just make sure you are passing the correct chain in - if it is working on the chain that is not moving, it will look like its reporting zero, but its not working on the correct chain....

Sun, 2016-12-11 10:53
jadolfbr