A BBC interview discussing the impact of these results (sepsis discussion begins @ 13:53):
The R scripts for running the multi-cohort analysis (data not included):
The R scripts for running the multi-cohort analysis (same as above, data not included):

The preprint linked below will be updated with the final, peer-reviewed version once published.

Sweeney, TE et al., "Mortality prediction in sepsis via gene expression analysis: a community approach". BioArxiv, 2016

The dataset GSE72829 described in Herberg et al, JAMA 2016 is composed of 4 smaller datasets spread across three microarray platforms. We used COCONUT to pool these data into a single dataset. Notably, since GSE80412 has no healthy controls, these were initially co-normalized using Limma with GSE72809, and then healthy controls between all three remaining datasets were used to make the final COCONUT-conormalized object. The pooled data are available as a .RDS (R object) here.

Some of the data used in the manuscript (the Inflammation and Host Response to Injury Glue Grant data) require IRB access. This can be obtained by following the instructions here:

Briefly, you will need to first register as a Consortium Member: Glue Grant Consortium Registration

You will then be able to access the consortium page, which explains exact instructions for obtaining data access at the bottom section, titled “Our Policies for Access to Data, Methodologies, and Protocols”.

When access to the Glue Grant database has been issued by the Glue Grant team, please email (1) IRB approval and (2) the letter of approval from the Glue Grant consortium to: tes17 [at] stanford [dot] edu to request access to the data object.

  • sepsis.txt
  • Last modified: 2018/09/05 13:46
  • by alexskr