PHYPred: a tool for identifying bacteriophage enzymes and hydrolases

Bacteriophages are viruses that attack bacteria and kill them through the lytic replication cycle. Many studies have reported that phages are more specific to bacteria than antibiotics are; thus, phage therapy has many potential applications in human medicine, with the advantage of having few side effects (Keen, 2012). Investigating the mechanisms of bacteria-killing phages will therefore aid in the development of antibacterial drugs. Hydrolases encoded by phages play a key role in the interaction between phages and host bacteria. These enzymes act on the bacterial cell wall to kill the host bacteria and then release progeny phages (Nielsen et al., 1999). Thus, correctly identifying the hydrolases encoded by phages can provide important clues for not only studying the lytic mechanism of the phage-bacteria system but also discovering potential antibacterial drugs. With the accumulation of proteomics data, various machine-learning methods have been applied to predict functional phage proteins. Sequritan et al. designed an artificial neural network (ANN)-based method to predict viral structural proteins using amino acid frequency (Seguritan et al., 2012). Recently, a special type of structural protein, phage virion protein, was identified using primary sequence information (Ding et al., 2014; Feng et al., 2013). However, to our knowledge, no computational method has been developed to predict phage hydrolases. Thus, the aim of this letter is to describe a powerful model for identifying phage hydrolases. We started by discrimi-nating phage enzymes from phage non-enzymes. Once a phage protein is recognized as phage enzyme, the model will determine whether the predicted enzyme is phage hydrolase. See it on Scoop.it, via Virology and Bioinformatics from Virology.ca
PHYPred: a tool for identifying bacteriophage enzymes and hydrolases
Source: Viral Bioinformatics