The use of sequence comparison to detect 'identities' in tRNA genes

We have developed a computational method that detects 'identities' in tRNA genes by using principal component analysis to classify the sequences of bases in tRNA genes into groups of similar sequences and then comparing the distribution of sequences of bases, in order to extract characteristic bases that are conserved within a group but differ between groups. These classification and comparison procedures are applied recursively to classify the sequences into hierarchical groups, so that multiple levels of characteristic sites can be detected. By using this computational method, we were able to detect many characteristic sites in the T and D domains of tRNAs, as well as the characteristic sites that had already been detected experimentally. This suggests that bases not only in the contact regions but also in the elbow regions, which determine the structure and dynamics of the whole tRNA molecule, are important to the tRNA-aminoacyl tRNA synthetase recognition.