Watch a real analysis form live

Differential expression was tested using DESeq2 Wald test on Genomarker 0.1.0, comparing two groups of sizes n_a=8, n_b=8. P-values were adjusted with the Benjamini-Hochberg (BH) procedure at alpha=0.05. 318 features were tested; 66 were significant after adjustment. Dataset fingerprint=74039024. Software: joblib=1.6.0, numpy=2.4.4, pandas=2.3.3, pikepdf=8.15.1, pydeseq2=0.5.4, rfc8785=0.1.4, scikit-learn=1.8.0, scipy=1.17.1, statsmodels=0.14.6, weasyprint=70.0, xgboost=2.1.4. Receipt: https://genomarker.com/api/receipts/d8e1159c-e8e6-51fc-bc05-ad318713b3a3. The fold-change threshold was folded into the test rather than applied afterwards (the null hypothesis is |log2 fold change| <= 1.0), so each adjusted p-value tests whether the effect exceeds that threshold and the false-discovery rate controls that hypothesis. Pre-analysis confounder screening was performed: surrogate-variable analysis (Buja-Eyuboglu permutation procedure, B=20) estimated 1 latent factor(s) (maximum |Spearman r| between an estimated surrogate variable and the case-control variable = 0.054); the strongest batch-condition association across the 1 designated technical variable(s) had Cramer's V = 0.000 (chi-square p = 1); the smallest batch had 8 samples; no data-quality issues were flagged.

a8e70863bdd70ad3…