Calculation of TM-score and rmsd for all predicted models by I-TASSER

[FYaghoobizadeh]

I submitted my job on the I-TASSER server (ID ) and get the results. According to the server help page, a higher c-score (i.e., more positive) is better. Is it true? accordingly between 5 top predicted models for my structure (with C-score -2.72, -2.32, -1.20, -3.39, and -2.94) I choose the model 3 with c-score -1.20. However, TM-score and RMSD are only provided for model 1 (C-score -2.72). In this regard, I validate the structures of model1 and model3 by saves v6.0 and proSA servers and get the better results for model 3.
How could we obtain these indexes for other models? Moreover, I compared the 3d of model3 with 10-top ranked model presented in I-TASSER results page with Swiss model structural assessment tool. Is this work to calculate TM-score and RMSD OK?
I will be very thankful if you guide me as rapid as possible.
Sincerely yours

[cenat]

Relying on just one model with complete scores while thenulls brawl ios others are missing key metrics makes it hard to trust the outcome fully. It feels like there’s a lot of guesswork involved, and I wish there were more balanced ways to evaluate all the models side by side.

[Terenobinson]

Yes — a higher (less negative / more positive) C-score is better.
To get TM-score and RMSD for the other models, download each model PDB from the I-TASSER results and run a structure-comparison tool (e.g. TM-align / TMscore) pairwise against your reference model. Example local command:
TMalign model3.pdb model1.pdb
This will print TM-score and RMSD. Using SAVES/ProSA for additional validation is fine — they give complementary quality checks.

[ningfieldin]

You’re right — higher (less negative) C-scores usually indicate more reliable models, but validating with ProSA and SAVES is a smart move since they give structural quality insights beyond the server’s rankingSprunki Game.