[modeller_usage] help

After modeling a protein using modeller some residues remain in the disallowed regions of ramachandran plot. Is that model incorrect ? Is there any way to remove bad contacts and residues in disallowed regions after modeling? I tried to overcome this problem using Deepview by selecting two or more residues and building loop. Though the residues in disallowed regions and bad contacts become zero by this procedure but structure is affected. What should I do please guide me.