I want to make protein structure for that protein DNAH17. For this purpose , I take protein sequence from UniProt:https://www.uniprot.org/uniprotkb/Q9UFH2/entry#sequences.
It contains almost 4500 aa so which modelling software i can use to predict structure.
I tried AlphaFold2.ipynb from google colab but it was not executed due to the big size of protein.
Should I try modeller?
It is almost a guarantee that you won't be able to model the whole protein except by using AlphaFold. Also, pretty much a guarantee that there is no free service out there that will let you model a ~4500 protein. A quick scan of PDB indicates that there is no template that covers even close to the whole protein. You should verify this last statement.
That leaves you with two options: find the domain organization of this protein and split it into smaller chunks, than model each of them separately with AlphaFold on Colab. You can try to model with Modeller, but you will most likely have to split the protein again since there is no continuous template for this protein. In case you don't know, Modeller can only do homology-based modeling (unlike AlphaFold, which doesn't require modeling templates). The latter option assumes you know how to use Modeller, which is not easy even for proteins of regular size.
The latter option assumes you know how to use Modeller
Do you know how to use Modeller? I vaguely recall there being a minimum homology threshold that needs to be met, and 63% sounds far from it, but it's ages since I used Modeller and my knowledge is definitely outdated.
How familiar are you with various protein structure modeling/prediction approaches?