Host respones and therapeutic interventions

Summary

We develop computational approaches to study host responses to infection and identify host-directed therapeutic interventions. We unified and standardized computational metrics for quantifying reversal of disease gene signatures by candidate therapeutics, addressing critical reproducibility issues in connectivity score methodologies. Our framework reconciles multiple scoring approaches (GSEA, KS, cosine, XSum) and provides quantitative guidance for method selection based on data characteristics.

Building on this, we developed an integrative drug repurposing pipeline for tuberculosis that combines transcriptomic signatures from diverse infection models with multi-database screening (LINCS L1000, CMap). Consensus approaches across multiple connectivity scores improve prediction of drugs with known anti-TB efficacy and identify novel repurposed candidates.