output interpretation

[nmrworker]

I just installed the stand-alone i-tasser 5.1 on FC 25 and finished the 1st peptide structure prediction. However, I don't know which files to look at, eg, which file corresponds to the "Predicted Secondary Structure", "Top 10 templates used by I-TASSER", "Top 5 Models predicted by I-TASSER", etc., because I can't easily tell by the output filenames. Also does the predicted structure contains all the protons? If not, is there an option to include them?

Thanks in advance.

[xiaogenz]

The predicted secondary structure can be found in "seq.psi.ss" and "seq.pss.ss". Top ranked templates used by I-TASSER can be found in "init.dat". "Top 5 Models predicted by I-TASSER" named as "REF_mode1.pdb" or "model1.pdb". The final model includes all atoms.

[nmrworker]

the "model1.pdb" generated by stand-alone version has ONLY the amide proton, but the one generated by the online server has all the protons. How to set the stand-alone version to generate all the protons?

[ITASSERteam]

you can download and run HAAD program (https://zhanglab.ccmb.med.umich.edu/HAAD/) to add all protons. It is very simple and fast.

[nmrworker]

Is that how the online service add protons to the structure? Also the sidechain orientations from the stand-alone version is quite different than the online version. Is that normal?

[robpearc]

1) No, as the standalone package uses Modrefiner to refine the protein conformation, while the online server uses a combination of several programs to perform the refinement. However, you may uncomment line 15 in the EMrefinement.pl included in the I-TASSER package, which will add hydrogens via HAAD.

2) Yes, the side-chain orientations may be different due to the use of 3rd party programs that we cannot include the downloadable version of I-TASSER.

[nmrworker]

It seems that you would recommend the online version over the stand-alone version since multiple programs are involved to improve the final results, is that right?

[robpearc]

Depending on your target, the results may be slightly better for the web server as it includes a few more threading methods that aren't included in the downloadable version. This will influence the results for more distantly homologous threading targets, where it is difficult to identify suitable templates from the PDB. However, for homology modeling, where there are obvious homologous templates, the results should be quite similar. In terms of the side-chain orientations, I would recommend you use our recently developed side-chain packing method, https://zhanglab.dcmb.med.umich.edu/FASPR/ to quickly repack the side-chains of your model, which was developed after the release of our latest I-TASSER package, so it wasn't included in it.

[matumbrella]

There is no direct option in I-TASSER to include all protons during the prediction process,dinosaur game as I-TASSER focuses on heavy atom positioning for peptide backbones and side chains. Therefore, post-processing with external software is necessary to include all protons.

[terassda]

Check the file named ss.txt in the output folder. This file contains the predicted secondary structure in a format such as H (helix), E (strand), and C (coil).
Look at the files named model1.pdb, model2.pdb, etc., where the top 5 models are saved as PDB files. These files are ranked based on their C-score (confidence score), and the C-scores are reported in the file model1.pdb.