Enrique Merino, Ph.D.

Research

The increasing number of publicly available whole-genome sequences opens opportunities to address new and important biological questions by comparative genomic approaches. We are particularly interested in identifying conserved regulatory motifs, without using any knowledge of regulon structure or metabolic pathways, so that general and unbiased searches can be performed. We have developed an easy and highly accurate operon prediction method based on intergenic distances of contiguous genes and the functional relationship scores of the STRING database between the different groups of orthologous proteins, as defined in the COG database. Nevertheless, the operon prediction of our method was not restricted to only those genes with a COG assignation. For an important number of genes that have not been annotated in the COG database, we successfully defined new groups of orthologous genes and obtained, by extrapolation, a set of equivalent STRING-like scores based on conserved gene pairs on different genomes. One of the most important advantages of our operon prediction method is the use of the very well know STRING scores that have been manly used to predict protein-protein interactions. Since the STRING functional relationships scores are determined in an un-bias manner and efficiently integrates a large amount of information coming from different sources and kind of evidences, the prediction made by our Neural Network are considerably less influenced by the bias imposed in the training procedure using one specific organism, and thus make it a suitable protocol for the operon predictions in the actual and future set of fully-sequenced genomes. (http://operons.ibt.unam.mx/OperonPredictor/)