As I understand you are first looking for a "diseased" protein to work upon and later it's structure prediction. So, this protein should have an implication directly in the progression of disease due to mutation (like in human cancer) or a pathogenic protein? Anyway searching in the PDB is a good start, you can read about them and their function. But that means the structure is already present in the database.
You have to search for a protein whose structure is yet not elucidated? I would suggest to do a little literature search. Find one human disease on which you would like to work to narrow down your search
Chetna Tyagi thanks for your suggestion i try the same procedure and find a protein involve in cancer and many other diseases its PDB structure is also not predicted