A temporal classifier predicts histopathology state and parses acute-chronic phasing in inflammatory bowel disease patients

We characterized the transcriptional landscape of acute and chronic models of dextran sodium sulfate (DSS) and adoptive transfer (AT) colitis murine models to derive novel temporal gene expression and splicing signatures in blood and colonic tissue in order to capture dynamics of colitis remission and relapse. We show that colitis induces robust expression of minor isoforms and using machine learning, demonstrate that specific temporal gene expression and splicing signatures are highly predictive of tissue histopathology spanning blood and colonic tissue as well as various experimental sites and protocols. Finally, we identified sub networks of patient-derived causal networks that were enriched in these temporal signatures to distinguish acute and chronic disease components within the broader molecular landscape of IBD. Overall design: Case-control differential gene-expression with temporal interaction