Methods for predicting bacterial protein subcellular localization

The prediction of a bacterial protein's subcellular localization can be of considerable aid to microbiological research. It can be used to infer potential functions for a protein, to either design or support the results of particular experimental approaches and, in the case of surface-exposed proteins, to quickly identify potential drug or vaccine targets in a given pathogen genome, or potential diagnostic/detection targets in pathogen or environmental isolates.Bacterial proteins contain sequence features that either directly influence the targeting of a protein to a particular cellular compartment or else are characteristic of proteins found at a specific localization site. These features are encoded in the protein's amino-acid sequence and can be identified computationally.By analyzing a protein for the presence or absence of one or more of these features and integrating the results, a prediction of which compartment a protein is likely to reside in can be generated.Since the 1991 release of the first comprehensive, web-based bacterial protein localization prediction method, PSORT I, seven other such tools have been released. This review summarizes the techniques implemented by each tool, their benefits, pitfalls and predictive performance.The review also describes alternative methods for localization prediction, including similarity searches against localization databases and the use of predictive tools designed to identify individual sequence features. The performance of these methods is compared with that of the seven broad-spectrum localization prediction tools.PSORTb and Proteome Analyst are the most precise predictive methods currently available, with other methods complementing them when higher sensitivity (a larger number of predictions) is required.The precision of certain localization prediction tools has now surpassed the precision of some high-throughput laboratory methods for localization determination. We can now reliably assign potential localization sites to the majority of proteins encoded in a genome.