Understanding Results
Learn how to interpret your Query Card and use the results in your research.
Query Card Anatomy
Every query generates a Query Card with these components:
1. Natural Language Answer
The top section provides a plain-English interpretation of the findings:
“DDR1 expression is significantly higher in papillary RCC (KIRP) compared to clear cell RCC (KIRC), with a 2.3-fold increase (p < 0.001).”
This summary is designed to be:
- Accessible to non-experts
- Accurate and statistically grounded
- Ready to share with collaborators
2. Interactive Figure
The visualization varies by analysis type:
| Analysis Type | Figure Type |
|---|---|
| Differential Expression | Boxplot with individual points |
| Tumor vs Normal | Grouped boxplot |
| Mutation Frequency | Bar chart |
| Survival Analysis | Kaplan-Meier curve |
Interactivity features:
- Hover: See exact values for data points
- Zoom: Click and drag to zoom into regions
- Pan: Shift+drag to move around
- Reset: Double-click to reset view
3. Statistical Results
Key statistics are displayed prominently:
All p-values are two-sided unless otherwise noted.
For expression comparisons:
- P-value: Statistical significance (typically Wilcoxon rank-sum test)
- Fold Change: Magnitude of difference between groups
- Sample Sizes: Number of patients in each group
- Effect Size: Cohen’s d or similar measure
For survival analysis:
- Log-rank P-value: Significance of survival difference
- Hazard Ratio: Risk of event in high vs low expression group
- Median Survival: Median time to event for each group
4. Methods Text
A grant-ready paragraph describing the analysis:
“Gene expression data were obtained from The Cancer Genome Atlas (TCGA). DDR1 mRNA expression (log2 TPM+1) was compared between papillary renal cell carcinoma (KIRP, n=290) and clear cell renal cell carcinoma (KIRC, n=533) using a Wilcoxon rank-sum test. All analyses were performed using BioQuery (bioquery.io).”
Click Copy Methods to copy this text for your paper.
5. SQL Query
For transparency and reproducibility, the exact SQL query is available:
SELECT sample_id, cancer_type, gene_expression
FROM tcga.expression
WHERE gene_symbol = 'DDR1'
AND cancer_type IN ('KIRP', 'KIRC')Click View SQL to see the full query.
Interpreting P-values
P-values should be interpreted in context. A small p-value with small effect size may not be biologically meaningful.
| P-value | Interpretation |
|---|---|
| p < 0.001 | Strong evidence against null hypothesis |
| p < 0.01 | Moderate evidence |
| p < 0.05 | Weak evidence (traditional threshold) |
| p ≥ 0.05 | Insufficient evidence |
Consider also:
- Sample size (large N can yield small p-values for trivial effects)
- Effect size (is the difference biologically meaningful?)
- Multiple comparisons (if running many queries, adjust significance threshold)
Interpreting Effect Sizes
Fold Change
For expression comparisons, fold change indicates magnitude:
| Fold Change | Interpretation |
|---|---|
| > 4 | Very large difference |
| 2-4 | Large difference |
| 1.5-2 | Moderate difference |
| 1-1.5 | Small difference |
Hazard Ratio
For survival analysis:
| Hazard Ratio | Interpretation |
|---|---|
| HR > 2 | Strong association with worse survival |
| HR 1.5-2 | Moderate association |
| HR 1-1.5 | Weak association |
| HR = 1 | No association |
| HR < 1 | Association with better survival |
Sample Size Considerations
Results with small sample sizes (n < 20 per group) should be interpreted cautiously.
BioQuery displays sample sizes prominently. Consider:
- TCGA data is observational - associations may not be causal
- Some cancer types have limited samples - rare cancers may have n < 50
- Subgroup analyses reduce effective sample size
Reproducibility
Every Query Card includes:
- Timestamp - When the query was run
- Data Version - Which TCGA data release
- SQL Query - Exact database query
- Statistical Methods - Tests and parameters used
This ensures your results can be reproduced and verified.