Experimentally Observed Quercetin Perturbations Show Context-Dependent Reversal-Compatible Correspondence with Human MDD Molecular States
Description
Background Major depressive disorder (MDD) comprises molecular states that vary across brain regions, sexes, cell populations, cohorts, and tissue compartments. Whether experimentally observed quercetin responses oppose these heterogeneous human MDD states remains uncertain. Methods We assembled 88 human MDD molecular-state signatures spanning brain and blood contexts and compared 11 experimentally observed quercetin signatures from nine human datasets with 87 eligible MDD signatures, yielding 957 prespecified pairs. Correspondence was evaluated separately using continuous gene effects, signed ranks, polarity-aware Top-200 overlaps, and Hallmark and Reactome pathway profiles. A 978-gene MCF7 UniPert–G2CP prediction was examined as auxiliary evidence and descriptively calibrated against observed quercetin contrasts. Context-matched non-quercetin comparators and cross-database protein-target reconstruction provided additional specificity and target-level context. Results Observed quercetin–MDD correspondence was heterogeneous across tissues, molecular resolutions, and evidence roles. Reversal-compatible and same-direction patterns coexisted across gene-level, Top-200, and pathway analyses, supporting bounded rather than universal molecular opposition. The G2CP prediction showed reversal-compatible correspondence with a restricted subset of MDD states, but predicted–observed calibration was predominantly near neutral across datasets. Comparator analyses showed context- and metric-dependent differences, while formal quercetin specificity remained unresolved. Twenty-eight human protein targets had cross-database support, but none met the predefined MDD brain transcriptomic criterion; downstream target-centered analyses were therefore not pursued. Conclusions Experimentally observed quercetin perturbations show context-dependent reversal-compatible correspondence with selected human MDD molecular states. These findings are hypothesis-generating and do not establish clinical antidepressant efficacy, human brain exposure, compound specificity, or direct target engagement. Separating observed perturbations from computational prediction, comparator calibration, and disease-constrained target evidence provides a transparent framework for evaluating natural compounds against heterogeneous psychiatric molecular states. Keywords: major depressive disorder; quercetin; transcriptomic perturbation; signature reversal; single-nucleus transcriptomics; virtual perturbation
Files
Steps to reproduce
1. Download and extract the public reproducibility package while preserving the original directory structure. 2. Read README.md and the accompanying file manifest before running the analysis. The manifest records the included files and their SHA-256 checksums. 3. Open the extracted project in R/RStudio and use the extracted package directory as the project root. 4. Install the required R packages listed in the README, scripts, or session-information file. 5. Review the path settings before execution. If a script contains an absolute local path (for example, D:/2026-quercetin/), replace it in a local working copy with the corresponding path inside the extracted package. The archived source files should not be modified. 6. Run the analysis scripts in the sequence documented in the README or according to their numerical prefixes. Intermediate and final outputs will be written to the corresponding output directories. 7. Compare the regenerated tables, figures, and result files with the supplied reference outputs. File integrity can be checked using MANIFEST_SHA256.txt and the public file manifest. The public package retains the latest identifiable version of each duplicated file. Byte-identical duplicates were collapsed. Ambiguous or methodologically distinct alternatives are retained separately in the full researcher bundle so that researchers may inspect and select them when performing sensitivity or alternative analyses. The combined document “supplementary materials 2026.8.23.docx” is intentionally excluded from both shared ZIP archives. Supplementary Tables S5–S12 are not part of this study and are therefore not expected.