Skip to content
Priyanshu Kumar
ReleasedZenodo · Version 1.0 · May 2026

SepsisTensor v1

A harmonized multi-cohort transcriptomic resource for mortality prediction

Dataset

Overview

Seven public sepsis cohorts covering 1,636 patients across microarray and RNA-seq platforms, combined into a shared 7,964-gene matrix using identifier mapping, cohort-wise standardization, and ComBat harmonization.

The release provides three analysis-ready tensors: a ComBat-corrected atlas of all cohorts, a training tensor of six cohorts, and an unharmonized holdout cohort kept separate for external validation.

GEO retrieval · Probe-to-gene mapping · Z-score standardization · ComBat

Cohorts
GSE185263, GSE236713, GSE26440, GSE272769, GSE54514, GSE65682, GSE95233
Patients
1,636 (367 non-survivors, 1,269 survivors)
Genes
7,964 shared across cohorts
Holdout
GSE65682 (n = 479), not harmonized

Figures

Four principal-component scatter plots: cohorts separated before harmonization, mixed after ComBat, and the mortality covariate retained.
Fig. 2 of the 36-gene study. Principal component analysis of the integrated atlas before (A) and after (B, C) ComBat harmonization, and the mortality covariate check (D).