Research Tools
Scientific Figure Studio — Publication-Quality Plots and Correlations
Runs in your browser
Example loaded.
Data
Open a CSV or Excel file, paste data, or start from an example. When you update the file, reload it to redraw the figure with the same settings.
Drop a CSV, TSV or Excel file here.
Used for the files you open.
Read 1.234,5 as 1234.5 (use semicolons between columns).
| Column | Type | Numbers | Missing | Not numbers |
|---|---|---|---|---|
| Site | text (3 groups) | 0 | 0 | 54 |
| Depth_cm | numeric | 54 | 0 | 0 |
| OrganicC_pct | numeric | 54 | 0 | 0 |
| TotalN_pct | numeric | 54 | 0 | 0 |
| Moisture_pct | numeric | 54 | 0 | 0 |
| pH | numeric | 54 | 0 | 0 |
54 rows. Empty cells and NA, N/A, NaN, null or - are treated as missing values.
Figure type
Relationship between two measurements, with regression lines.
Encodings
Which columns drive position, colour, size and shape.
Text: groups. Numbers: a colour scale.
Y columns
Statistics
Equation, R², correlation, p and n above the plot.
Axes and legend
Use ^{ } for superscripts and _{ } for subscripts, e.g. mmol L^{-1} or CO_{2}.
Reference lines and labels
Thresholds (e.g. a detection limit or y = 0) and text placed in data units.
Size and style
Column widths and font sizes from publishers’ artwork guidelines; check your journal’s current instructions.
For numeric colour columns and heatmaps.
Colour of each group
Figure
89 × 68 mm · Helvetica/Arial 7 pt · 54 values
Export
Vector SVG and PDF for print; PNG and TIFF with the resolution written into the file.
2,102 × 1,606 pixels
Settings files store every choice (columns by name), so you can redraw the same figure after updating your data. Nothing is saved in your browser.
Correlations
Every pair of numeric columns, strongest first. Rows with a missing value are left out of that pair only.
Linear relationships; 95% CI by Fisher’s z.
10 pairs tested.
Columns to correlate
None chosen: all numeric columns are used.
| Variables | r | 95% CI | p | n | Strength | Plot |
|---|---|---|---|---|---|---|
| OrganicC_pct × TotalN_pct | 0.99*** | 0.99 to 1.00 | < 0.001 | 54 | strong positive | |
| OrganicC_pct × Moisture_pct | 0.91*** | 0.84 to 0.94 | < 0.001 | 54 | strong positive | |
| OrganicC_pct × pH | −0.90*** | −0.94 to −0.84 | < 0.001 | 54 | strong negative | |
| TotalN_pct × Moisture_pct | 0.90*** | 0.83 to 0.94 | < 0.001 | 54 | strong positive | |
| TotalN_pct × pH | −0.89*** | −0.94 to −0.82 | < 0.001 | 54 | strong negative | |
| Depth_cm × TotalN_pct | −0.87*** | −0.93 to −0.79 | < 0.001 | 54 | strong negative | |
| Depth_cm × OrganicC_pct | −0.87*** | −0.92 to −0.79 | < 0.001 | 54 | strong negative | |
| Depth_cm × pH | 0.84*** | 0.74 to 0.91 | < 0.001 | 54 | strong positive | |
| Moisture_pct × pH | −0.78*** | −0.87 to −0.65 | < 0.001 | 54 | strong negative | |
| Depth_cm × Moisture_pct | −0.75*** | −0.85 to −0.61 | < 0.001 | 54 | strong negative |
Stars: * p < 0.05, ** p < 0.01, *** p < 0.001. Strength follows common rules of thumb (|r| ≥ 0.7 strong, ≥ 0.4 moderate, ≥ 0.1 weak). Correlation does not imply causation, and many tests on the same data will find some “significant” pairs by chance, which the adjustment controls.
Summary statistics
For each series or group in the figure. SD uses n − 1; the 95% CI of the mean uses the t distribution.
| Group | n | Mean | SD | SEM | 95% CI | Median | IQR | Min | Max |
|---|---|---|---|---|---|---|---|---|---|
| Site A | 18 | 0.9889 | 0.8918 | 0.2102 | 0.5454 to 1.432 | 0.885 | 0.15 to 1.647 | 0.15 | 2.73 |
| Site B | 18 | 1.448 | 1.191 | 0.2807 | 0.8561 to 2.041 | 1.57 | 0.175 to 2.518 | 0.15 | 3.43 |
| Site C | 18 | 0.7456 | 0.5218 | 0.123 | 0.4861 to 1.005 | 0.685 | 0.2575 to 1.195 | 0.15 | 1.61 |
How to use this tool
- 1Open a CSV or Excel file, paste data, or choose an example; check the text encoding and the column types.
- 2Choose the figure type and map columns to the x and y axes, colour, size and shape.
- 3Set the statistics, axes, reference lines, labels and a journal size preset.
- 4Review the correlations and summary statistics, then export PDF, SVG, PNG or TIFF.
Frequently asked questions
Scatter plots (with linear regression, a 95% confidence band and the equation, R², r, p and n), line plots (every row, or the mean or median of replicates with SEM, SD or 95% CI as error bars or a band), bar charts (mean, median, sum or count with error bars and the individual points), box plots, violin plots, histograms (count, percent or density on shared bins), correlation heatmaps and scatter matrices.
An encoding links a column to a visual property. Besides the x and y axes you can map a text column to colour groups, a numeric column to a colour scale or to point size, and a text column to marker shape. Several y columns can be drawn together. Your file is also read in its own text encoding: UTF-8 and UTF-16 are recognised automatically, legacy code pages such as Windows-1252, Windows-1250 or Shift_JIS are detected, and you can choose another if letters look wrong.
Pearson’s r (with a 95% confidence interval from Fisher’s z), Spearman’s ρ and Kendall’s τ-b are calculated for every pair of numeric columns, using the rows where both values are present. p-values use the t distribution (Pearson, Spearman) or the tie-corrected normal approximation (Kendall), and can be adjusted for multiple comparisons with Holm’s method or the Benjamini–Hochberg false discovery rate. The results are checked against SciPy and statsmodels in the tool’s tests.
Presets set the width, height, font and line weights for Nature (89 and 183 mm), Science (55 and 120 mm), Cell Press (85 and 174 mm), Elsevier (90 and 190 mm), IEEE (3.5 and 7.16 in), PLOS (132 mm), a thesis page, a presentation slide and a poster panel. Figures are drawn in true millimetres and point sizes, so a 7 pt label is 7 pt when printed at 100%. Guidelines change, so check your journal’s current instructions.
Use the PDF or SVG for vector artwork; text and lines stay sharp at any size and can be edited in Illustrator, Inkscape or Affinity Designer. For journals that ask for images, download a TIFF or PNG at 300, 600 or 1200 dpi; the resolution is written into the file so submission systems read it correctly.
In titles, axis labels, reference-line labels and text labels, write ^{…} for a superscript and _{…} for a subscript, for example mmol L^{-1}, R^{2} or CO_{2}. Column names and group names from your data are always printed exactly as written.
No. Settings refer to columns by name. Open the updated file (or use Reload file) and the figure is redrawn with the same settings when the columns match. You can also save the settings as a JSON file and load them later with new data.
No. Reading the file, the statistics and the figure are all done in your browser; large files are parsed in a background worker on your device. Nothing is uploaded or stored.
Related tools
CSV to PGFPlots — Scientific LaTeX Graph Generator
Turn CSV data into a PGFPlots figure for LaTeX: line, scatter, bar, grouped and horizontal bar charts, histograms and box plots, with error bars, log axes and a live preview. Download a .tex file that compiles as it is, or a ZIP with the data as a separate CSV.
CSV & Excel to LaTeX Table Generator and Editor
Import CSV, TSV or Excel data, edit it in a spreadsheet grid, and generate publication-quality LaTeX tables: booktabs, longtable, tabularx and rotated tables, with captions, labels, column groups, notes and exact decimal formatting.
PRISMA 2020 Flow Diagram Generator for Systematic Reviews
Enter the records, reports and studies at each stage of your systematic review and get a PRISMA 2020 flow diagram in the official layout, for new or updated reviews, with or without other sources. Counts are checked branch by branch, and the diagram exports as SVG, PNG or PDF.
BibTeX & BibLaTeX Cleaner, Validator and Formatter
Clean a .bib file without breaking it. Normalise indentation, field order, DOIs, months and page ranges; find duplicate keys and DOIs, empty, unknown and missing fields, bad URLs and syntax errors; review every change side by side and accept or reject it before downloading.