Using Co-expression Mode

Co-expression mode colors the embedding using the combined expression of multiple genes. Use it to find cells where a group of genes is expressed, compare their expression patterns, or see where two genes vary in the same or opposite directions.

Plotting multiple genes

1. On the Molecular Data page, click the Plot gene expression button next to a gene. The embedding switches to show that gene's expression.
2. Select Co-expression above the embedding.
3. Add more genes using the buttons in the gene panel. You can also hold Ctrl, Command, or Shift while clicking a gene's plot button to add it directly to co-expression mode.

The selected genes are listed above the embedding. Click the remove button next to a gene to remove it from the plot. You can plot up to six genes at once.

To return to the expression view for a single gene, select Single. The first gene in the co-expression list remains plotted.

Choosing a co-expression mode

Use the Co-expression menu above the embedding to choose how the selected genes are combined:

• Highest Value: Colors each cell using the highest expression value among the selected genes. This highlights cells where any selected gene is strongly expressed while preserving expression intensity.
• Expressed in All: Assigns a value of 1 when every selected gene has expression above zero in a cell, and 0 otherwise. Use this to identify cells that express all selected genes.
• Expressed in Any: Assigns a value of 1 when at least one selected gene has expression above zero in a cell, and 0 otherwise. Use this to identify cells that express one or more selected genes.
• Correlation: Compares exactly two genes and shows whether their expression deviates from their dataset-wide means in the same or opposite directions for each cell.

For cell i, the Correlation score is calculated as:

score_i = ((x_i - mean(x)) * (y_i - mean(y)))
          / max_j(abs((x_j - mean(x)) * (y_j - mean(y))))

Here, x_i and y_i are the expression values of the two genes in cell i. The means and the maximum used for normalization are calculated across all cells in the dataset.

• A score close to 1 represents a strong same-direction deviation: both genes are above their means or both are below their means.
• A score close to -1 represents a strong opposite-direction deviation: one gene is above its mean while the other is below its mean.
• A score of 0 means at least one gene is at its mean. Scores near 0 indicate weak co-deviation.