Energy Calculation in Rosetta3.4

Member Site Forums Rosetta 3 Rosetta 3 – General Energy Calculation in Rosetta3.4

Viewing 6 reply threads
  • Author
    Posts
    • #1233
      Anonymous

        I have about 100000 PDB files and I need to calculate their free energy and total energy.

        Does Rosetta3.4 have any such tool to calculate it ?

        Please advise.

      • #6963
        Anonymous

          Depending on how you define “free energy” and “total energy”, this is easy. Rosetta does not calculate in physical energy units (kcal/mol), it calculates in Rosetta Energy Units, which don’t directly convert to physical energies. Read some of Kellog and Baker’s papers on DDGs to learn how they’ve tried to get a scorefunction that reports in kcal/mol.

          Anyway, if you’ll take Rosetta’s scores as answers, then make a file containing the filenames of all your pdbs (like, say “ls *pdb > pdblist”), and pass it to the score_jd2 app with the -l flag:

          score_jd2.release -database /path/to/database -l pdblist -jd2:delete_old_poses

          For 100,000, scoring is fast, but you may want to use MPI if it’s available. (You need delete_old_poses to prevent a memory leak that occurs in this situation)

        • #6968
          Anonymous

            I am getting a scoring file (score.sc) which looks like

            SCORE: score fa_atr fa_rep fa_sol fa_intra_rep pro_close fa_pair hbond_sr_bb hbond_lr_bb hbond_bb_sc hbond_sc dslf_ss_dst dslf_cs_ang dslf_ss_dih dslf_ca_dih rama omega fa_dun p_aa_pp ref linear_chainbreak overlap_chainbreak time description
            SCORE: 2877.394 -266.363 3036.076 123.882 2.540 0.000 -2.674 -19.786 -16.727 -3.889 -1.492 0.000 0.000 0.000 0.000 -6.296 4.673 44.258 -9.408 -7.400 0.000 0.000 1.000 S_00000001_0001_0001
            SCORE: 3615.740 -295.051 3789.284 142.326 2.541 0.000 -5.800 -19.779 -16.738 -4.709 -2.170 0.000 0.000 0.000 0.000 -6.300 4.668 44.277 -9.409 -7.400 0.000 0.000 0.000 S_00000001_0002_0001
            SCORE: 464.803 -262.327 623.065 118.856 2.543 0.000 -2.091 -19.783 -16.736 -3.251 -1.320

            Which part of this is generally used as the energy of the pdb molecule ?

          • #6969
            Anonymous

              The first column, marked “score”, is the weighted sum of the other terms, and is used as the overall energy.

              Those are scores that indicate Rosetta hates the structures, BTW – you should have something like -2 per residue, you have something positive. The positivity is in your fullatom repulsive (fa_rep) term.

            • #6970
              Anonymous

                It is a dimer so may be that is why it is positive.

              • #6971
                Anonymous

                  No…if there are 100 residues total, you should expect fullatom scores between -100 to -300, depending on the quality of the structure. A positive score means Rosetta things the structure has a positive folding energy – in other words it won’t fold that way. (Usually this is due to a clash).

                • #6972
                  Anonymous

                    Keep in mind that “clashes” are *very* common in structures which come directly from the PDB. Rosetta’s energy function is sensitive enough that small coordinate changes which aren’t necessarily experimentally relevant can have a great effect on the reported energy. (And even a small clash in some distal part of the protein can greatly affect the score.)

                    It’s commonly suggested that from-PDB structures should be relaxed into the Rosetta energy function before relying on the score. The recently-released Rosetta 3.4 now has a simple protocol to do a heavy-atom constrained relax, which in our experience can greatly improve the energy without significantly changing the atom positions (i.e. all-atom RMSDs in the tenths of angstoms). ( See http://www.rosettacommons.org/manuals/archive/rosetta3.4_user_guide/d9/d2e/prepare_pdb_for_rosetta_with_relax.html, sections labeled with “Relax with all-heavy-atom constraints”)

                Viewing 6 reply threads
                • You must be logged in to reply to this topic.