A clean, client-side visual tool to compare two differential expression signatures across omics.
No server, no database, no uploads stored. Everything runs on your machine.
Technical reference for OmicLine decisions and definitions.
log2FC quantifies relative change between conditions. A value of 1 means a twofold increase; −1 means a twofold decrease. Values near zero indicate no meaningful change.
The adjusted p-value (padj) corrects for multiple testing, typically via the Benjamini-Hochberg procedure. OmicLine uses this as the significance threshold alongside log2FC to define differentially expressed genes.
Combines magnitude and significance into a single metric:
padj is floored at 1×10⁻¹⁰ before computation to prevent score inflation from extremely small p-values. This floor is applied silently and logged in the session file.
A gene is considered a DEG (differentially expressed gene) if:
Default cutoffs: |log2FC| ≥ 1 and padj ≤ 0.05. Both must be met within a single condition.
Pearson correlation and visualization are computed over genes that are DEGs in at least one of the two conditions (OR logic). Genes that fail the DEG filter in both conditions are excluded from the plot and from the correlation.
Pearson r measures linear concordance between log2FC values of the two conditions within the analysis universe. Values near +1 indicate strong concordance; values near −1 indicate inversion.
The default ranking is radial distance from the origin:
This prioritizes genes with large effect sizes in either or both conditions. Top 200 genes are shown by default; the maximum is 3000.
By default, the regression line is forced through the origin (intercept = 0). An option for free intercept is available in the advanced panel. The forced-zero model is appropriate when comparing the same type of measurement and expecting proportional scaling.
Axis limits are set at the 99th percentile of |log2FC| values to prevent extreme outliers from compressing the plot. Truncated points are still present in the data and in exports. A banner appears if truncation occurs.
When a gene symbol appears multiple times in a single table (e.g., multiple probes), OmicLine keeps the row with the highest absolute integrated score. The event is logged.
OmicLine does not reprocess raw data, normalize expression values, correct for inverted contrasts, compare more than two tables, or impute missing genes. It is a visualization and concordance tool for pre-computed differential results.