Query the evidence
Search every harmonized evidence record — all genes in all reanalysed study contrasts, including nonsignificant results — not just the ten displayed findings per study. The full table (~17 MB) downloads into your browser once; every query then runs locally, and nothing you type leaves the page.
These are study-level molecular associations, not validated biomarkers. Higher expression does not necessarily mean greater resilience, rows that share controls are not independent replication, and no cross-study pooled effect is computed. Rows marked exploratory come from the tank-dependent PRJNA735889 analysis: their p-values are nominal only. See how expression findings relate to resilience and the guide to reading results.
| Gene | Description | Study | Contrast | Tissue | Stage | Stressor | Effect (log2FC) | SE | Adj. p | Interpretation | Quality flags |
|---|
Positive effects indicate higher expression in the treatment relative to the control. Adjusted p-values from the exploratory PRJNA735889 contrast are nominal only. Hover a truncated cell for the full text; follow a study link for its design, replication and limitations.
Advanced: run SQL against the evidence table
Columns match the harmonized evidence schema:
study_id, bioproject, feature_id_standardized, description, sample_comparison, effect_size,
effect_size_type, standard_error, p_value, adjusted_p_value, sample_size, tissue, life_stage,
stressor, phenotype, resilience_classification, inference_status, quality_flags.
Prefer a script or a chatbot? The same file powers both: download aree_evidence.parquet and query it with DuckDB, pandas or polars, or hand its URL to an AI assistant. Full per-study TSVs with complete provenance columns are linked from real-study findings.