B2R research figures — data and methods
All data is synthetic, seed 20261010; no experimental, patient, or client records.

Dose-response plate: 3 series × 9 concentrations × 4 synthetic replicates = 108 observations. Concentration is in micromolar on a logarithmic axis. Unweighted least-squares four-parameter inhibitory logistic fit: Bottom + (Top-Bottom)/(1+(c/IC50)^Hill). IC50 locates the midpoint between fitted plateaus, not necessarily an absolute 50% response. Error bars are sample SD; residuals are observed minus fitted response in percentage points. No confidence interval or efficacy conclusion is claimed.

Expression plate: 400 abstract features and 6 synthetic samples per group. Input is a synthetic normalized log2 expression matrix, NOT RNA-seq counts. Two-sided Welch t-tests with BH adjustment. Log2 fold change is mean(B)-mean(A); colored points meet adjusted P<0.05 and |log2FC|≥1. PCA uses all 400 features, centered by feature without variance scaling; explained variance is computed by SVD. The heatmap uses the 24 highest-variance features, row Z-scores using sample SD, average-linkage Euclidean row clustering, and columns in fixed group order. Color scale is clipped at ±2.5. No DESeq2, UMAP, or enrichment analysis was run; there are no real gene or pathway findings.

Specialist bioinformatics requests require a feasibility assessment. These figures demonstrate visual presentation, not validated specialist service capability.