Online PCA Analysis and Plot

Paste a gene expression matrix to run principal component analysis (PCA) in your browser, color samples by group, label them, and export SVG / PNG plus the sample scores.

Everything runs locally in your browser. Pasted data is never uploaded to any server.

1. Expression matrix

Choose file

2. Sample groups (optional)

One group name per sample in column order (one per line, or comma-separated), or one “sample name + Tab + group” per line. Leave empty for no groups.

Group colors

3. Preprocessing

log2(x + 1) transformTop N genes by variance (0 = all)Standardize each gene (center + scale, platform default)

4. Plot

X axisY axisPoint size (pt)Width (mm)Title

Label samples95% confidence ellipse per group

Paste your data, or click “Load example” to try it out.

Same method as the platform: the 1000 genes with the highest variance, each centered and divided by its standard deviation (R prcomp(center = TRUE, scale. = TRUE)); variance explained matches prcomp. The sign of a PC is arbitrary; here the sample with the largest absolute score on each PC is made positive.
Normalized, log-scale input is recommended (e.g. DESeq2 VST, log2 CPM / TPM). Raw counts should not go into PCA directly; the platform applies VST first, and without DESeq2 here, log2(x + 1) is offered as an approximation.
The 95% ellipses follow ggplot2 stat_ellipse defaults (multivariate t, robust estimate) and need at least 4 samples per group. With few replicates they are large and only indicative.
Rows with missing values are removed, genes with identical values in all samples are dropped first, up to 500 samples are supported, and with many samples the number of genes is capped automatically (2000 genes at 500 samples) to keep the browser responsive. Layout follows the platform figure style (Arial, 6 pt text); width 89 or 183 mm.