Estimating the relative contributions of virulence factors for pathogenic microbes.

Many pathogenic microbes have multiple virulence factors that can cause damage to the host and thus contribute to an overall virulence phenotype for that organism. Although current techniques are suitable for demonstrating that a particular microbial characteristic contributes to virulence, no formal approach for defining the relative contributions of multiple virulence factors to overall virulence has been proposed. This paper describes the use of multivariate linear regression to estimate the relative contributions of virulence factors to the overall phenomenon of virulence. The approach is illustrated here with sample calculations of the relative contributions of individual Cryptococcus neoformans and Bacillus anthracis virulence factors to the overall virulence phenotype. These calculations were derived from a small underpowered experimental data set for the fungus and two larger sets of randomly generated data for both microbes. The major limitation of this method is a requirement for large data sets of microbial strains that differ in virulence and virulence factor expression. Multivariate linear regression can be used to identify the relative levels of importance of virulence factors in virulence studies, and this information can be used to prioritize antigen identification for vaccine development and the design of antimicrobial strategies that target virulence mechanisms.