Genome-wide survey of protein families and superfamilies

The availability of whole genome information of several model organisms offers the possibility to perform analyses of protein families to decipher accurate protein function prediction. Association of gene products to pre-existing families is a popular approach to the prediction of function. However, despite the availability of huge amounts of valuable protein sequence information, nearly half of the gene products cannot be attributed structure or function by direct methods, obviating the intervention of bioinformatics inputs. In this review, we survey early whole-genome computational surveys of model organisms performed using bioinformatics algorithms such as fold recognition methods and ultrasensitive sequence search procedures. Fold recognition methods apply inverse folding approaches to recognize the fold of gene products. Heuristic sequence search algorithms, like PSI-BLAST, employ rapid but sensitive strategies to recognize relationships of gene products to pre-existing folds. We also discuss applications and analyses that are possible starting from such genome-wide surveys.