ab initio structure modeling

dear rosetta users.

i’ve been trying to predict a structure of protein. it is 76aa in length. and using experiment, it is demonstrated that it has very compact  structure.

i think that it is ideal to predict structure, using rosetta. however. unexpectedly, i think that predicted structure is not perfect.

i used Ab initio application of rosetta3.12 to create about 20000 decoys. and because of no nature crystal structure, i selected one decoy of predicted decoys with the lowest total_score as nature structure. and then i rescored the 20000 decoys using score_jd2 -in:file:silent fold.out in:file:native S_00000212_60. and i drawed the total_score_vs_rms plot. the plot didn’t have

funnel structures. moreover, the rescored total_score and rms are much greater than i expected.

is there any way to deal with this problems.

thanks in advance.

P.S. the related experiment related with this protein shows that it has 5 disulifide bonds. isn’t it useful in predicted the correct structure?

it is flag file for ab initio prediction

-in

        -file

                -fasta ./halC.fasta       # protein sequence in fasta format

                -frag3 ./aat000_03_05.200_v1_3    # protein 3-residue fragments file

                -frag9 ./aat000_09_05.200_v1_3    # protein 9-residue fragments file

-psipred_ss2 ./t000_.psipred_ss2

-nstruct 300

-abinitio

        -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.

-run

        -protocol broker

        -reinitialize_mover_for_each_job # jd generate fresh copy of its mover before each apply (once per job)

-score

        -find_neighbors_3dgrid  # Use a 3D lookup table for doing neighbor calculations. For spherical, well-distributed conformations

-out

        -file

                -silent ./fold_silent_${SLURM_ARRAY_TASK_ID}.out # full path to silent file output

-overwrite      # overwrite any existing output with the same name you may have generated