I want to build Homology model of a protein whose sequence length is 3390 amino acids. Please let me know is it possible to build the homology model of such a large protein and if so how far we can rely on that?
You can model bits of it - the subdomains and possibly some combinations of them if they are covered by a single template structure.
Such large proteins typically have large regions of disorder - it's well worth using a disorder predictor as part of the process to determine what regions can be modelled.
Whether you can rely on the model really depends on what your question is. If you are interested in a structured region which is covered by a template you may be able to get something useful.
It's not obvious what you mean by 'work' - see my above comment, whether you get a satisfactory answer depends completely on why you are making the model.
The general rule is you can model the individual domains up to quite a high level of accuracy if there are good template structures which have been conserved structurally - either owing to a short evolutionary distance or high selective constraint.
Combinations of domains will not be easy to model unless there is a structure which covers the combination; additionally there is likely to be some flexibility in how the domains interact which means you will at best get one possible answer.
The best answer I can give you without knowing anything about *which* protein you are trying to model is that it will be very difficult, you're not likely to get an accurate structure of the whole thing even without disordered regions being considered.
I was also thinking of That sir g, and I also found large disordered region ~690 Amino acids. I have submitted the sequence to different servers I am waiting what are they going to do.