# Rosetta ab initio run gives 0.00 RMSD for generated 2000 models? (run still in progress)

6 posts / 0 new
Rosetta ab initio run gives 0.00 RMSD for generated 2000 models? (run still in progress)
#1

Dear Friends,

I am running ab initio rosetta run using openmpi:

mpiexec -np 16 /main/source/bin/AbinitioRelax.mpi.linuxgccrelease -database /main/database/ @options -mpi_tracer_to_file log > log1.log &

I see "score.fsc" file in the folder with 2000 models, and generated a RMSD vs Energy plot, whcih gives me this (the attached plot)

All 2000 models have 0.00 RMSD, is this normal? Do I need to wait for the models to converge? your commensts and solutions are very welcome. thanks

AttachmentSize
28.3 KB
Post Situation:
Mon, 2019-09-23 03:24
Danielsebas

Hello,

I cannot be sure without seeing your 'options' file, but it looks like to me you haven't added the flag:

-in:file:native {my_native_pdb_file}

which gives rosetta a structure to calculate an RMSD against.

Mon, 2019-09-23 07:29
danpf

Thanks! there is no template pdb as it is a de novo structure prediciton. My options file is:

# Make sure all variable names have been replaced with absolute path and that no line begins with a \$ or ~s
-in
-file
-fasta sequence.fasta         # protein sequence in fasta format
-frag3 t001_.200.3mers  # protein 3-residue fragments file
-frag9 t001_.200.9mers  # protein 9-residue fragments file
-abinitio
-relax
-increase_cycles 10     # Increase the number of cycles at each stage in AbinitioRelax by this factor
-rg_reweight 0.5        # Reweight contribution of radius of gyration to total score by this scale factor
-rsd_wt_helix 0.5       # Reweight env, pair, and cb scores for helix residues by this factor
-rsd_wt_loop 0.5        # Reweight env, pair, and cb scores for loop residues by this factor
-relax
-fast   # At the end of the de novo protein_folding, do a relax step of type "FastRelax".  This has been shown to be the best deal for speed and robustness.
-out
-nstruct 50000  # how many structures do you want to generate?  Usually want to fold at least 1,000.
-file
-silent abrelax.out # full path to silent file output
-silent_struct_type binary      # we want binary silent files
-scorefile score.sc
-overwrite      # overwrite any existing output with the same name you may have generated
-nstruct 50000


Mon, 2019-09-23 07:47
Danielsebas

Rosetta can't give an RMSD if you don't give it a structure to RMSD against.  RMSD is relevant when you have a "correct structure" to compare to (meaning, for benchmarking purposes).  If you are doing an unknown protein it's not a useful metric here.

After generating models you can do a second-pass re-scoring calculating RMSD against the best model.  If that score versus RMSD plot shows a funnel it's a good sign that the model is correct.

Mon, 2019-09-23 09:24
smlewis

thanks! Could you please let me know how to do this re-scoring? i am trying to find it in the tutorials but not successfull. thanks

Mon, 2019-09-23 09:43
Danielsebas

I don't remember, but off the top of my head, just run them all through score_jd2 with -in:file:native set to whichever model you want them compared to.

Mon, 2019-09-23 11:42
smlewis