Combining algorithms to predict Bacterial protein sub-cellular location: Parallel versus concurrent implementations

Paul D Taylor,Darren R Flower,Teresa K Attwood

doi:10.6026/97320630001285

Abstract

We describe a novel and potentially important tool for candidate subunit vaccine selection through in silico reverse-vaccinology. A set of Bayesian networks able to make individual predictions for specific subcellular locations is implemented in three pipelines with different architectures: a parallel implementation with a confidence level-based decision engine and two serial implementations with a hierarchical decision structure, one initially rooted by prediction between membrane types and another rooted by soluble versus membrane prediction. The parallel pipeline outperformed the serial pipeline, but took twice as long to execute. The soluble-rooted serial pipeline outperformed the membrane-rooted predictor. Assessment using genomic test sets was more equivocal, as many more predictions are made by the parallel pipeline, yet the serial pipeline identifies 22 more of the 74 proteins of known location.

Highlights

Subcellular location is amongst the prime arbiters of host immunogenicity within Bacteria
The parallel pipeline outperforms the individual methods for three of the compartments, which are all of interest to vaccinologists: Gram-positive extracellular proteins (11.24% more accurate), Gram-negative extracellular proteins (3.58% more accurate) and outer membrane proteins (8.04% more accurate)
The combination method learns from such discombobulatingly similar amino acid compositions. correlations, increasing the capacity of the network to Conclusion: The principle purpose of in silico reverse-vaccinology is to identify potential vaccine targets, but it is important to reduce significantly the number of targets to be tested by successfully removing intracellular proteins

Summary

Introduction

Subcellular location is amongst the prime arbiters of host immunogenicity within Bacteria. For a successful - and practical - in silico reverse-vaccinology analysis we need to predict multiple localisations reliably and consistent accuracy. A viable strategy for achieving this aim is to combine together a set of individual binary predictors, which discriminate between a positive and a negative class, and develop a functional reverse-vaccinology pipeline. There are two common implementations of pipelines for prediction of properties from data: parallel and serial. The multiple modules that are to be combined are run simultaneously and a further new module is required to decide which module has made the correct prediction. Employ a tree-like method for the execution of the individual binary modules. To make a classification with a serial pipeline, the root module is used first to analyse the data and Bioinformation 1(8): 285-289 (2006)

Objectives

Methods

Findings

Conclusion