ddg_monomer: How the ddG is actually calculated based on the 50 trajectories? What scorefunction is used?

[Anonymous]

(relevant files can be downloaded at https://copy.com/yOhxRslZzK8j9z0l)

Dear friends,
In the instructions about ddg_monomer,

https://www.rosettacommons.org/docs/latest/ddg-monomer.html

it is mentioned that “the most accurate ddG is taken as the difference between the mean of the top-3-scoring wild type structures and the top-3-scoring point-mutant structures”.

In my “ddg_predictions.out”, the “total” term seems to indicate a ddG of 1.247.

However, after using score_jd2.linuxgccrelease to score the “wt_.out” and “mut_A111C.out”, “total_score” and “score” are shown for all the 50 trajectories. Afterwards, I rank those 50 trajectories in “wt_.out” and “mut_A111C.out”, and calculate “the difference between the mean of the top-3-scoring wild type structures and the top-3-scoring point-mutant structures”. What I got is:

For “total_score”:
(-723.7-723.59-723.301)/3-(-724.728-723.934-723.709)/3=0.593

For “score”:
(-1096.21-1096.104-1095.652)/3-(-1097.457-1096.407-1096.143)/3=0.680

As you can see, neither 0.593 nor 0.680 is identical to 1.247.

So can I ask what is the exact scorefunction ddg_monomer uses to calculate the “total” term in “ddg_predictions.out”?

Thank you very much.

Yours sincerely
Cheng

[Anonymous]

Your ddg_monomer scoring function appears to have been talaris2013, To go into detail, your log output displays the following scoring terms and weights (for an example state).

---

Scores Weight Raw Score Wghtd.Score

---

fa_atr 0.800 -2823.941 -2259.152
fa_rep 0.440 543.884 239.309
fa_sol 0.750 1545.485 1159.114
fa_intra_rep 0.004 828.621 3.314
fa_elec 0.700 -249.860 -174.902
pro_close 1.000 1.635 1.635
hbond_sr_bb 1.170 -34.251 -40.073
hbond_lr_bb 1.170 -144.095 -168.591
hbond_bb_sc 1.170 -45.162 -52.839
hbond_sc 1.100 -31.661 -34.827
dslf_fa13 1.000 -3.033 -3.033
atom_pair_constraint 1.000 8.692 8.692
rama 0.200 -102.002 -20.400
omega 0.500 64.018 32.009
fa_dun 0.560 549.834 307.907
p_aa_pp 0.320 -228.629 -73.161
ref 1.000 25.170 25.170

---

Total weighted score: -1049.829

Looking at your scoring output, you were also using talaris2013 for scoring. Notably, the final lines of log output are

protocols.moves.ddGMover: mutate A111C wildtype_dG is: -1097.46 and mutant_dG is: -1096.21 ddG is: 1.24653
apps.public.ddg.ddg_monomer: mutation label for this round is A111C

What was your command line? I suspect that whatever your flags were, it just took the difference of the two best-scoring structures, rather than the paper protocol of the average of the best three.

[Anonymous]

> Is there something incorrect if I want to use talaris2013 while keeping other parts the same as the paper protocol?

Not necessarily; you’d just asked.

> In addtion, are you sure the ~-1000 score is for talaris2013 in my protocol?

Looking at your ddg log output, I see a ton of lines outputting scores around -1000. I suspect that the score line in your scoring output is similarly the score under talaris. I am not sure what total_score is referring to here or why it differs from score; I would ignore it for now (I’ll keep thinking).

[Anonymous]

Hi everyday847,
Thank you for your help. The attachments are my command line and options file. Is there something incorrect if I want to use talaris2013 while keeping other parts the same as the paper protocol?

In addtion, are you sure the ~-1000 score is for talaris2013 in my protocol? The thing is, when I use score_jd2 to score the structure, I only got score of ~-770 and I think talaris2013 is default for score_jd2.

Thank you.

Yours sincerely
Cheng

#main body for the command line:
ddg_monomer.linuxgccrelease @ /home/ucbechz/Scratch/20150203_ddg_monomer_8398mutants/input/options/A111C.options > ddg.log

#options file:
-ddg::mut_file /home/ucbechz/Scratch/20150203_ddg_monomer_8398mutants/input/mutfile/A111C.mutfile
-in:file:s /home/ucbechz/Scratch/20150203_ddg_monomer_8398mutants/input/C226S.pdb
-constraints::cst_file /home/ucbechz/Scratch/20150203_ddg_monomer_8398mutants/input/input.cst
-ddg:weight_file soft_rep_design
-ddg:minimization_scorefunction talaris2013
-ddg::iterations 50
-ddg::dump_pdbs true
-ignore_unrecognized_res
-ddg::local_opt_only false
-ddg::min_cst true
-ddg::suppress_checkpointing true
-in::file::fullatom
-ddg::mean false
-ddg::min true
-ddg::sc_min_only false
-ddg::ramp_repulsive true
-unmute core.optimization.LineMinimizer
-ddg::output_silent true
-override_rsd_type_limit

[Anonymous]

Hi everyday847,
Thank you. I look forward to your new thought about the score.

Yours sincerely
Cheng

[Author]

Posts