A renormalization approach is introduced in quantum thermal annealing. This new global optimization algorithm (QTAR) is applied to a highly frustrated BLN model protein with 46 residues. The underlying quantum sampling algorithms utilized are primitive and staging path integral Monte Carlo. The current method enables us to achieve significant improvement in the success rate for locating the global minimum of this protein, while using much less computational effort, compared to previous schemes used. Further applications and possible enhancements of the algorithm are also discussed.