Many previous attempts to derive survival models in whole blood transcriptomics data have failed. Statistical corrections of blood transcriptomics data are needed for multiple comparisons. However, these corrections reduce sensitivity and therefore can constrain discovery. Additionally, signals of inflammation in deep organs may be very weak and hard to detect in the blood, since the blood is not the site of perturbation. On the other hand, although signals in affected tissue are likely much stronger, deep tissues are much less accessible than peripheral blood. To overcome these limitations and allow discovery in peripheral blood, we hypothesized that transcriptional signatures from affected tissues can aid in blood transcriptomics analysis. The aim was to predict death in mice infected with lethal vs. non-lethal doses of influenza virus by examining gene expression in peripheral blood. SOURCE: Andrew,James,Martins (andrew.martins@nih.gov) - Systems Genomics and Bioinformatics National Institute of Allergy and Infectious Diseases

Pluto Bioinformatics

GSE124404: Blood RNA-seq of mice after influenza infection