A Comparison of Computational Methods for Identifying Virulence Factors

Abstract

Bacterial pathogens continue to threaten public health worldwide today. Identification of bacterial virulence factors can help to find novel drug/vaccine targets against pathogenicity. It can also help to reveal the mechanisms of the related diseases at the molecular level. With the explosive growth in protein sequences generated in the postgenomic age, it is highly desired to develop computational methods for rapidly and effectively identifying virulence factors according to their sequence information alone. In this study, based on the protein-protein interaction networks from the STRING database, a novel network-based method was proposed for identifying the virulence factors in the proteomes of UPEC 536, UPEC CFT073, P. aeruginosa PAO1, L. pneumophila Philadelphia 1, C. jejuni NCTC 11168 and M. tuberculosis H37Rv. Evaluated on the same benchmark datasets derived from the aforementioned species, the identification accuracies achieved by the network-based method were around 0.9, significantly higher than those by the sequence-based methods such as BLAST, feature selection and VirulentPred. Further analysis showed that the functional associations such as the gene neighborhood and co-occurrence were the primary associations between these virulence factors in the STRING database. The high success rates indicate that the network-based method is quite promising. The novel approach holds high potential for identifying virulence factors in many other various organisms as well because it can be easily extended to identify the virulence factors in many other bacterial species, as long as the relevant significant statistical data are available for them.

Novel bacterial ADP-ribosylating toxins: structure and function

“…While bioinformatics has been used to identify other putative virulence factors 16 , the number of available data sets for each group of virulence factors is limited when compared to the number of identified bARTT sequences and available crystal structures 186 . However, although the ability to predict a bARTT is strongly supported, this predictive capability wanes when he aim is to identify the substrate.…”

Typing methods based on whole genome sequencing data

“…92.44.117/virulent/index.html). Differently from the previously described sequence based training, some publicly unavailable approaches rely on classification algorithms utilizing sequence associated information from biological repositories such as gene ontology, functional domains and protein-protein network information [79][80][81]. A recent review concludes that ML-based virulence prediction methods frequently perform worse than BLAST-similarity based approaches [77].…”

Virulence factor activity relationships (VFARs): a bioinformatics perspective

“…These include virulent-GO, 49 integrated query networks, 50 and protein–protein interaction networks. 51 Virulent-GO is a Gene Ontology (GO) annotation-based tool for the prediction of virulent proteins in bacterial pathogens. 49 GO annotation describes the functions of genes and their products across the species.…”

“…Protein–protein interaction (PPI) networks are based on Search Tool for the Retrieval of Interactive Genes/Proteins database for the identification of virulence factors in the proteomes of different bacteria. 51 In PPI networks, two types of information are considered for the prediction of virulence factors: (i) the number of neighbouring nodes (to proteins), and (ii) confidence scores based on strengths of interactions with them. The identification accuracy of PPI networks is around 0.9–significantly higher than those obtained using sequence-based prediction methods.…”