Turn numeric variables into a publication-ready correlation heatmap.
Upload CSV or Excel data, pick Pearson or Spearman correlation, and SciFigure computes every pairwise coefficient before generating a clean, labeled matrix for your manuscript.
Upload data and describe the matrix you need.
Costs 5 credits. Coefficients are computed from your uploaded data, then drawn as a PNG.
Preview
Aspect ratio: 16:9
These are sample previews. Upload your dataset to generate your own correlation matrix.
Use these examples as starting points for biomarker panels, gene co-expression, environmental measurements, and clinical variable screens.
Prompt
Based on the uploaded biomarker_panel.csv file, create a Pearson correlation matrix for CRP, IL6, TNFa, BMI, Age, Glucose, and HDL. Use a diverging blue-white-red scale from -1 to +1, print r values with two decimals, add a labeled colorbar, and keep the full symmetric matrix.
Prompt
Based on the uploaded gene_coexpression.xlsx file, create a clustered Pearson correlation matrix for 10 genes. Order variables so strongly correlated genes sit together, outline the proliferation and p53-response blocks, omit numbers in cells, and use a compact colorbar.
Prompt
Based on the uploaded soil_chemistry.csv file, create a Spearman rank correlation matrix for pH, organic carbon, nitrogen, phosphorus, moisture, and microbial biomass. Show the lower triangle only, print rho values, and use an earthy but colorblind-safe diverging palette.
Prompt
Based on the uploaded clinical_variables.tsv file, create a lower-triangle Pearson correlation matrix for systolic BP, diastolic BP, LDL, HDL, HbA1c, BMI, and age. Use circle size and color to encode r, print values above |0.3|, and keep labels angled for readability.
Prompt
Based on the uploaded wellbeing_scales.csv file, create a Pearson correlation matrix for nine questionnaire subscales: perceived stress, anxiety, depression, sleep quality, rumination, self-efficacy, social support, mindfulness, and life satisfaction. Show the upper triangle only, print r with two decimals, and add significance asterisks with a footnote explaining them. Use a blue-white-red scale from -1 to +1 and rotate the column labels 45 degrees.
Prompt
Based on the uploaded station_climate.xlsx file, create a Pearson correlation matrix for mean temperature, precipitation, relative humidity, wind speed, solar radiation, and evapotranspiration measured at 48 weather stations. Render it in grayscale for a print-only journal, running from white at -1 through mid gray at 0 to near black at +1, print r in every cell, outline each cell in black, and keep the full symmetric matrix with horizontal labels.
Add CSV, TSV, TXT, or Excel data with one row per sample and one numeric column per variable. ID columns are excluded automatically.
Choose Pearson or Spearman, full or lower triangle, clustered or original order, and whether r values appear in each cell.
SciFigure computes every coefficient, draws the heatmap, and lets you download a PNG ready for papers and slides.
It computes the correlation coefficient for every pair of numeric variables in your table and draws the results as a color-coded matrix. Positive relationships, negative relationships, and near-zero associations become visible at a glance, which makes the figure useful for exploratory analysis, multicollinearity checks, biomarker screening, and supplementary figures in research papers. Coefficients are calculated from your uploaded values before any drawing happens, so the colors and printed numbers reflect your data rather than an AI guess.
Use a wide table: one row per sample, subject, or site, and one numeric column per variable. Text columns such as sample IDs or group labels are excluded from the calculation.
The coefficients are calculated from your uploaded values using pairwise complete observations. The computed numbers are then passed to the image model, which only handles layout and styling.
Yes. Include "Spearman" in your instructions and the tool ranks each variable before computing correlations. Pearson is used by default.
Each pair of variables uses every row where both values are present, so one missing measurement does not remove the whole sample from the matrix.
Yes. Ask for clustered ordering to group strongly related variables, and ask for a lower-triangle layout to remove the duplicated upper half.
Yes, but verify the coefficients, method, sample size, and variable names before submission, and report whether p-values or multiple-testing corrections were applied in your own analysis.
Generate correlation matrices, heatmaps, PCA plots, and other research-ready visualizations with SciFigure AI.
