Ab initio prediction of multidomain protein structure

My predicted protein structure gives a C-score of -2.73 and TM-score 0.40±0.13. I would like to try to predict my protein by partitioning its domains as this may improve C-score. I never did that before and so I would just like to know how should I do it.

1- I have predicted my protein domain via ThreadomEX, it gives me three domains as follow
(1:135, 136:464, 465:602)

2-Should I take each predicted domain and submit it independently to AIDA web tool ?

3-Then should I download PDB file for each one from AIDA. I have also a problem that AIDA asks me to upload PDB file ?!!

4-Then should I assemble upload PDB files of these domains through https://zhanglab.ccmb.med.umich.edu/reindex_pdb/

So just to make sure I understand well, I have nothing to do with I-TASSER in this Ab-initio prediction assembly steps. Please let me know the correct steps in details to model my protein by partitioning its domains

Many Thanks for your time and consideration.