This panel shows the frequencies of cell types, cell type interactions, or Cell Neighborhoods across all the regions in the cohort. On the right, the rows are labeled with each of the regions. On the bottom, the columns are labeled with each of the cell types, cell type interactions, or Cell Neighborhoods.

The columns are ordered by a hierarchical clustering, with cell types/interactions/Cell Neighborhoods that have similar distributions among the regions grouped together. By default, the rows are separated by cohort, and a grey line divides the cohorts from each other. If you switch Row Clustering on, the regions will also be grouped by hierarchical clustering, with regions that have similar distributions of cell types/interactions/Cell Neighborhoods grouped. In this view, the bar on the left indicates which cohort each region belongs to, with cohort 1 colored blue, and cohort 2 colored orange.

Hover over a colored rectangle to see the data underlying it.

Other parameters you can customize:

• Normalization Method: How should the color scale in the plot be normalized? The options are:

  • Min-max: maximum values are scaled to 1, and the minimum values are scaled to 0, using the equation shown below.

    

  • Z-score: standard deviations away from the mean value

  • None: raw values

• Normalize By: By which dimension should the colors in the heatmap be scaled? The options are:
  • Row scaling: This is best for determining what the dominant cell type/interaction/Cell Neighborhood is in each region
  • Column scaling: This is best for determining which region have the largest proportion of each cell type/interaction/Cell Neighborhood

Downloading the heatmap and raw data

In the upper right corner of this panel, there are also buttons to allow you to download an SVG or PNG image of the heatmap, or a CSV file of the raw data underlying the heatmap.

Next:

Interaction Chord Diagram

Return to Home Directory