The overall goal of SCRIPT Research Project 2 is to create a computational model based on microbial biosignatures that predicts clinical failure in patients with hospital-acquired pneumonia. Specific pathogens such as Pseudomonas aeruginosa and Acinetobacter baumannii are particularly problematic in ventilator-associated pneumonia and are associated with clinical failure rates as high as 50%, even in patients treated with appropriate antibiotic therapy. For this reason, a more detailed analysis will be performed on pneumonia caused by these pathogens.
Project 2 will focus on two main areas: 1) the bacterial genomic profiles and 2) the microbiome changes and communities (including bacteria, bacteriophage, other viruses, and fungi) associated with unsuccessful outcomes in patients with ventilator-associated pneumonia.
Weighted correlation network analysis applied to shotgun DNA sequencing derived BAL-assigned species identifies modules of co-abundant microbiome factors associated with clinical features including pathogen culture. Top left: unsupervised clustering of species- level abundance is performed and clusters are assigned to modules designated by colors. Top right: to illustrate the approach, the purple module is highlighted. Bottom left: a strong, highly significant correlation is shown between the purple module membership and pathogen culture from BAL (Spearman correlation). Bottom right: purple module members have significantly more relatively abundance in the Pseudomonas culture positive BAL samples than other pathogens.