parapertussis infection in human populations, and our results suggest that concurrent B. pertussis infection may do the same. However, as far as we know, B. parapertussis infections have not emerged at high levels in the era of pertussis vaccine use, although diagnostics for B. parapertussis infections need to be improved before the picture is clear. Coinfection with these two closely related pathogens may be more common than documented in human pertussis disease and the less virulent of the pair may benefit from the immunomodulatory properties of B. pertussis. Of course, whether this mouse model is representative of human infection is

unclear. Some aspects of B. parapertussis infection in mice more closely resemble those of B. bronchiseptica than B. pertussis (Heininger et al., 2002), and it is possible that B. pertussis is better adapted to the human host Tanespimycin mw than B. parapertussis and would outcompete

it in a mixed infection in a MS 275 human. Human volunteer experiments may be necessary to resolve these issues. This work was supported by NIH grant AI063080. We thank Galina Artamonova and Aakanksha Pant for conducting some of the preliminary mouse infection studies and Charlotte Mitchell for technical advice with BAL. “
“Vaccines are very effective at preventing infectious disease but not all recipients mount a protective immune response to vaccination. Recently, gene expression profiles of PBMC samples in vaccinated individuals have been used to predict the development of protective immunity. However, the magnitude of change in gene expression that separates vaccine responders and nonresponders is likely to be small and distributed across networks of genes, making the selection of predictive and biologically relevant genes difficult.

Here we apply a new approach to predicting vaccine response based on coordinated upregulation of sets of biologically informative genes in postvaccination gene expression profiles. We found GPX6 that enrichment of gene sets related to proliferation and immunoglobulin genes accurately segregated high responders to influenza vaccination from low responders and achieved a prediction accuracy of 88% in an independent clinical trial. Many of the genes in these gene sets would not have been identified using conventional, single-gene level approaches because of their subtle upregulation in vaccine responders. Our results demonstrate that gene set enrichment method can capture subtle transcriptional changes and may be a generally useful approach for developing and interpreting predictive models of the human immune response. Vaccination is one of the most effective methods of preventing human disease. However, many vaccines are not universally protective and even widely used vaccines, such as those against influenza, fail to achieve protective immunity in a significant proportion of vaccinated subjects [1].