Automated Predictions by a Factor Graph GO Annotation in Pacu

MARIANO TORRES MANNO; FLAVIO SPETALE; PILAR BULACIO; JOAQUIN EZPELETA; TAPIA ELISABETH; SILVIA ARRANZ; GABRIELA VANINA VILLANOVA; FLAVIA KRSTICEVIC

Buenos Aires

Simposio; 2nd ISCB Latin AmericanStudentCouncilSymposium; 2016

International Society for computacional biology

Gene Ontology is a hierarchical controlled vocabulary used to describe gene function. Protein-coding genes are generally annotated through standard methods that commonly rely on sequence similarity or protein signature searches. Most of the gene functional characterization in model organisms remains without a complete GO annotation. The worst-case scenario is for non-model organisms, where genome reference is not available and gene novelties with unknown function are expected. As a consequence, alternative computational method for automated gene annotation can be useful. In this work a classification method based on factor graph GO annotation (FGGA) is considered to enrich the coding gene annotation in non-model organisms. Concerning characterization methods of individual protein sequences in terms of a fixed number of input features, the presence/absence of 453 of the physicochemical type and four of the secondary structure type, were considered. The aim of this work was evaluate the ability of FGGA predictions to obtain more specific GO terms in a set of 441 pair paralogous from the transcriptome in the non-model organism Pacu (Piaractus mesopotamicus). To recover most specific and confident FGGA predictions a cut threshold of 0.95 for leaf nodes was set for the analysis predicted graphs. Additionally, to extend biological knowledge in the Pacu, 5 genes with unknown function were considered. Finally, particular attention was given to specific genes involved in growth and metabolic process in Pacu. That target genes could be useful for markers identification for future breeding program.