[Prediction of functionally related proteins by comparative genomics in silico].

Review is devoted to computational prediction of functionally related proteins by comparative genomics. Growing possibilities of biotechnology for genome sequencing lead to generation of sequences for millions of genes. However, function of majority of these genes is unknown, and can be determined experimentally only for a few of them. Therefore, accurate and robust methods for in silico prediction (annotation) of gene functions are highly required. We describe here the main techniques of comparative genomics, including the standard method based on transferring functions between homologous sequences and also context-based methods, including phylogenetic profiles and gene-neighbor approaches. Modern methods of comparative genomics allow obtaining correct functional annotations for more than a half of all organism proteins.