The GINZU Algorithm

Domains that have a homologous protein for which a structure is known are better modeled using homology-modeling software. PSI-BLAST is used to find sequence homologues with a known structure. This is done by searching against sequences in numerous different databases, such as the Protein Data Bank and Pfam.

After PSI-BLAST, GINZU submits the part of the sequence that did not match a structure to a threading/fold recognition server (structure-structure comparison algorithm, ORFeus). Then parts of the sequence that did not match a sequence with a known structure are modeled by an ab initio algorithm. Our lab has developed one of the best ab initio algorithms available today, called Rosetta.

GINZU produces the predicted domain structure of the target sequence. Each of the domains will have a functional annotation, assigned different confidences.

CASP || GINZU || Rosetta || Examples