Zhanglab forum

I am confused about the use of D-I-TASSER for predicting protein structures with no close homology and evolutionarily correlated sequences. In such cases, the predicted structure has numerous errors in its Bond angles, steric clashes, and other aspects. Am I using the wrong tool? If so, what is the alternative and better tool for such purposes, or at least refining the structure?

There seems to be an issue in the final step of structure refinement, which we are currently investigating and working to resolve. In the meantime, you may use FG-MD (https://zhanglab.comp.nus.edu.sg/FG-MD/) or ModRefiner (https://zhanglab.comp.nus.edu.sg/ModRefiner/) to correct local structural clashes. Thank you.

ITASSERteam wrote: ↑Sun Jun 08, 2025 2:49 pm There seems to be an issue in the final step of structure refinement, which we are currently investigating and working to resolve. In the meantime, you may use FG-MD (https://zhanglab.comp.nus.edu.sg/FG-MD/) or ModRefiner (https://zhanglab.comp.nus.edu.sg/ModRefiner/) to correct local structural clashes. Thank you.

Thank you very much for your response, Please let me know if possible when the issue might gets resolved!

ITASSERteam wrote: ↑Sun Jun 08, 2025 2:49 pm There seems to be an issue in the final step of structure refinement, which we are currently investigating and working to resolve. In the meantime, you may use FG-MD (https://zhanglab.comp.nus.edu.sg/FG-MD/) or ModRefiner (https://zhanglab.comp.nus.edu.sg/ModRefiner/) to correct local structural clashes. Thank you.

I have tried ModRefiner as it supported better for moving forward without template, but still the refined structure has very low score for steric clashes, and the structure predicted is a lot different than what DITASSER and Alphafold did earlier.

Any suggestions?

Abd_Compbio wrote: ↑Fri Jun 13, 2025 1:26 pm

ITASSERteam wrote: ↑Sun Jun 08, 2025 2:49 pm There seems to be an issue in the final step of structure refinement, which we are currently investigating and working to resolve. In the meantime, you may use FG-MD (https://zhanglab.comp.nus.edu.sg/FG-MD/) or ModRefiner (https://zhanglab.comp.nus.edu.sg/ModRefiner/) to correct local structural clashes. Thank you.

I have tried ModRefiner as it supported better for moving forward without template, but still the refined structure has very low score for steric clashes, and the structure predicted is a lot different than what DITASSER and Alphafold did earlier.

Any suggestions?

We have fixed one of the key steps in FG-MD, please try to use FG-MD (https://zhanglab.comp.nus.edu.sg/FG-MD/) to refinement your protein monomer structure. We are testing the D-I-TASSER. When it is ready, we will tell you. Thanks!

junh wrote: ↑Sat Jun 14, 2025 1:21 am

Abd_Compbio wrote: ↑Fri Jun 13, 2025 1:26 pm

ITASSERteam wrote: ↑Sun Jun 08, 2025 2:49 pm There seems to be an issue in the final step of structure refinement, which we are currently investigating and working to resolve. In the meantime, you may use FG-MD (https://zhanglab.comp.nus.edu.sg/FG-MD/) or ModRefiner (https://zhanglab.comp.nus.edu.sg/ModRefiner/) to correct local structural clashes. Thank you.

I have tried ModRefiner as it supported better for moving forward without template, but still the refined structure has very low score for steric clashes, and the structure predicted is a lot different than what DITASSER and Alphafold did earlier.

Any suggestions?

We have fixed one of the key steps in FG-MD, please try to use FG-MD (https://zhanglab.comp.nus.edu.sg/FG-MD/) to refinement your protein monomer structure. We are testing the D-I-TASSER. When it is ready, we will tell you. Thanks!

I appreciate your help!

junh wrote: ↑Sat Jun 14, 2025 1:21 am

Abd_Compbio wrote: ↑Fri Jun 13, 2025 1:26 pm

ITASSERteam wrote: ↑Sun Jun 08, 2025 2:49 pm There seems to be an issue in the final step of structure refinement, which we are currently investigating and working to resolve. In the meantime, you may use FG-MD (https://zhanglab.comp.nus.edu.sg/FG-MD/) or ModRefiner (https://zhanglab.comp.nus.edu.sg/ModRefiner/) to correct local structural clashes. Thank you.

I have tried ModRefiner as it supported better for moving forward without template, but still the refined structure has very low score for steric clashes, and the structure predicted is a lot different than what DITASSER and Alphafold did earlier.

Any suggestions?

We have fixed one of the key steps in FG-MD, please try to use FG-MD (https://zhanglab.comp.nus.edu.sg/FG-MD/) to refinement your protein monomer structure. We are testing the D-I-TASSER. When it is ready, we will tell you. Thanks!

The FG-MD program is not run correctly in D-I-TASSER. Now we fix it. Please try it again. Or you can directly use the FG-MD server (https://zhanggroup.org/FG-MD) to remove the clashes.