Reproducibility and cross-dataset concordance of drug-induced transcriptional responses

Cross-source concordance of drug-induced transcriptional changes across six drug-perturbation transcriptomic resources is generally limited, even when drugs and conditions are matched.

• Why it matters: Users need reproducible, transferable drug-response signals from expanding perturbation datasets, but agreement in treated expression may not reflect repeatable drug-induced changes.
• What they did: The study compared six resources using independent replicates, matched-drug analyses, and response decomposition, including a Tahoe baseline using control-expression-only to assess unresolved drug-specific variation.
• The result: Drug main effects (average responses across cell lines) predict cell-specific interactions and this relationship held in cross-plate validation, while other exploratory structure/target/external-response signals were modest.

AI summary of the abstract · may contain errors · not medical advice · read the paper.