Modeling Tertiary Structure of Complementarity Determining Region of Antibodies

Knowing the tertiary structure of proteins is important to predict protein function and to achieve rational drug design. However, huge amounts of money and time are needed to determine tertiary structures by experimental method. Only a small number of protein structures have been experimentally determined; therefore, prediction of the tertiary structure of proteins by computer should be of great help. Here we report the prediction of the tertiary structure of antibodies, especially the complementarity determining region (CDR). As one example of such predictions the prediction of the CDR is particularly interesting for two reasons: the CDR determines the functionality of the antibody, and the CDR is very flexible, so determining its structure experimentally is difficult. CDR consists of six loops (H1-3, L1-3). The amino acid sequence of CDRs varies exceptionally largely, so homology modeling, which is the main prediction method conventionally used, is not useful. In a previous work, it was discovered that CDRs could be classified into several canonical structures [2]. There, the relations between amino acid sequence and canonical structures (called “rules”) were suggested. In this research, we investigated a method which predicts tertiary structures of CDRs based on the canonical structure rule. This method consists of mainly two parts: one is identifying the CDR, the other is finding CDRs having similar structures using the CDR rules.