You are here

Warnings during relax

3 posts / 0 new
Last post
Warnings during relax
#1

I am getting  a lot of warnings like this:

core.pose.util: [ WARNING ] Unable to find atom_tree atom for this Rosetta branch connection angle: residue 53 BRANCH 1
core.pose.util: [ WARNING ] Unable to find atom_tree atom for this Rosetta branch connection angle: residue 92 BRANCH 1
core.optimization.Minimizer: [ WARNING ] LBFGS MAX CYCLES 200 EXCEEDED, BUT FUNC NOT CONVERGED!
 

This is using the following relax command:

relax.static.linuxgccrelease -s "structure.pdb" -use_input_sc \
    -constrain_relax_to_start_coords -ignore_unrecognized_res \
    -nstruct 20 \
    -relax:coord_constrain_sidechains  \
    -relax:cartesian -score:weights ref2015_cart \
    -relax:min_type lbfgs_armijo_nonmonotone \
    -relax:script cart2.script

 

I am using Rosetta 3.10. Should I ignore these warnings?

Category: 
Post Situation: 
Thu, 2018-10-11 14:42
cossio

Hi,

I also encoutered the same WARNING while I was running the Example1B from the Tutorial: Rosetta tools  for  structure determination  in  cryoEM density.

core.pose.util: [ WARNING ] Unable to find atom_tree atom for this Rosetta branch connection angle: residue 32 BRANCH 1
core.pose.util: [ WARNING ] Unable to find atom_tree atom for this Rosetta branch connection angle: residue 74 BRANCH 1
core.pose.util: [ WARNING ] Unable to find atom_tree atom for this Rosetta branch connection angle: residue 225 BRANCH 1

I am using Rosetta 3.9.

I am not sure about the meaning of this warning.

Is it something I have to worried about in the future? 

Thanks a lot for your help

Best,

Martina

Thu, 2019-10-10 10:16
maudagnotto

The "Unable to find atom_tree atom for this Rosetta branch connection angle: residue " can appear because certain protocols (like RosettaCM) "chop up" the protein into smaller pieces. Even if the protocol doesn't do it, depending on the input, the way the input is layed out can result in "odd" connectivities of residues.

This normally isn't an issue, but in certain situations various protocols will attempt to change a degree of freedom which goes across these breaks. When that happens, you get the warning, as it's attempting to treat something as if it's bonded when it mechanistically isn't. For most protocols, this isn't an issue. Rosetta attempts to change that degree of freedom, but doesn't as it doesn't exist (leaving the structure unchanged), and then moves on with the protocol.  This might reduce the efficiency of the protocol a bit, but often doesn't substantially impact the results seen.

I wouldn't worry about it unless you're not getting decent results, in which case it might be a suggestion to look at how your structure is layed out and connected. (Though that might not actually do anything to improve your results.)

 

The "LBFGS MAX CYCLES 200 EXCEEDED, BUT FUNC NOT CONVERGED!" message is also not something to be concerned about, especially in a protocol like relax which has repeated application of the minimizer. Even if the minimizer isn't converged in that particular iteration, it probably got pretty close to convergence, and it'll have an opportunity to get closer on the next round when the minimizer is applied again.

Mon, 2019-11-04 13:56
rmoretti