ABSTRACT
Comparative modeling allows someone to build a three-dimensional model for a protein of known amino acid sequence, but unknown structure, using another protein of known sequence and structure as a template. Comparative modeling is also known as 'homology modeling', implying that models are always generated from homologous proteins. This chapter presents two major approaches to comparative modeling: 'traditional' method; and 'simulated annealing' approach. It discusses the major factors which contribute to the accuracy of a model and evaluates which regions are likely to be accurately modeled and which regions will be poor. The chapter describes the traditional method for building a protein model. It provides brief overview of MODELLER, a comparative modeling package from Andrej Sali which uses a simulated annealing approach and has been shown to do consistently well in the Critical Assessment of Structure Prediction (CASP) experiments.
