You are here

what rotamer lib is used in rosetta3.5?

7 posts / 0 new
Last post
what rotamer lib is used in rosetta3.5?

The release note shows Dunbrack 2010 library is added. But what is the default lib? Is 2002 or 2010? Thank you!

Post Situation: 
Thu, 2014-02-06 16:11

The default rotamer library for Rosetta3.5 is the 2002 Dunbrack library. Academic users can use the 2010 library instead by passing the -dun10 flag on the commandline.

For the weekly releases, though, concomitant with the switch from score12 to talaris2013, the default rotamer library was switched to Dunbrack 2010. The 2002 library should be available with the "-dun10 false" option setting, or, for preference, the "-restore_pre_talaris_2013_behavior" flag (which does other things besides just switching the rotamer library).

Fri, 2014-02-07 08:53

Thanks. Do we need to change the weight file of energy terms if change from 2002 to 2010? There is Score12Dun10 in "Scientific Benchmarks for Guiding Macromolecular Energy Function
Improvement". Which weight file should I choose in Rosetta3.5? Thank you!

Sun, 2014-02-09 16:50

If one were to use dun10 with Rosetta3.5 (as opposed to switching to the weeklies and talaris2013), what's suggested in the Methods in Enzymology paper would be what I would recommend using.

Mon, 2014-02-10 07:42

the weekly release help for score.linuxgccrelease shows no default value (setting) for -corrections:score:dun10 (type is B). Does that imply it is true by default? Or should one set it to true?

Option | Setting |Type| Description
corrections:score: | | |
dun10 | | B| Use the 2010 Dunbrack
| | | library instead of either
| | | the the 2002 library.
| | |

Thu, 2014-03-27 08:26

just for the record: in weekly build,

  • By default dun10 is used.
  • core.pack.dunbrack: Dunbrack 2010 library took 0.2 seconds to load from binary

  • To turn it on explicitly, use -corrections:score:dun10 true.
  • To turn it off, using -corrections:score:dun10 false. But talaris2013 is optimized for dun10(?). So turning dun10 off may be a bad idea(?)
  • If no boolean value is specified, like: -corrections:score:dun10, default value is used (in this case dun10).
Fri, 2014-03-28 02:58

How we treat the options system in Rosetta is a bit ... complicated. Sometimes things aren't a simple on/off flag value. We can test if a user explicitly set an option, so for some boolean options there's actually three settings: true, false, and "not explicitly specified" and due to historical reasons the behavior can be different between each of the three. There's also "meta" options which will implicitly set other options; often that's only if those other options haven't explicitly been set on the commandline, but not always. And then protocols can be free to use or not use options as they see fit - sometimes you'll see protocols which will observe option values under certain circumstances, but in other circumstances won't even bother to check them, regardless of whether they were explicitly set or not.

But yes, for the (academic*) weekly releases dun10 is used by default, unless you explicitly turn it off. But as you mention, talaris2013 is optimized for dun10, so I normally wouldn't recommend changing -dun10 by itself. Instead, if you need dun02 for some reason use the meta option "-restore_pre_talaris_2013_behavior", which will change a number of setting back to their old values, including turning dun10 off and changing the default scorefunction back to score12.

(* There was a licensing issue with the 2010 Dunbrack library and commercial use. That's now been addressed, so recent weeklies for commercial users should have the dun10 library, but some of earlier versions for commercial users may not have the dun10 libraries included.)

Fri, 2014-03-28 09:03