Characterizing the location of one cell phenotype in relation to another is one approach to investigate spatial patterns in a tissue. In this vein, the goal of the Spatial Neighbor Distance Extension is to determine the spatial proximity of two cell types by calculating the distance between them. Some examples of the utility of Spatial Neighbor Distance are computing the distance between Tumor cells to the nearest Endothelial cell, or computing the distance between CD8+ T cells to the nearest Dendritic cell.
<aside> 💡 Spatial Neighbor Distance computes the Euclidean distance from “Cell Type A” to the nearest “Cell Type B”.
</aside>
Spatial neighbor distance computes the Euclidean distance from “Cell Type A” to the nearest “Cell Type B” for all cells of “Cell Type A”.
Illustration of how distance measurements are computed: Euclidean distance from Cell Type A to nearest Cell Type B.
This measurement is distinct from pairwise analysis which analyzes cell-cell interactions, as “Cell Type A” and “Cell Type B” are not required to be in proximity of each other to measure the distance between them.
Read more about the underlying algorithm here.
The output of a Spatial Neighbor Distance run is a density plot of the distribution of distances for all cells of “Cell Type A” to the nearest “Cell Type B” for a given tissue region. The summary statistics for the generated density plots can be downloaded in .csv format. Currently, the results of these calculations cannot be aggregated or stratified by metadata features in the Portal. To determine whether the average distance between two cell types varies across cohorts, we recommend leveraging the summary statistics of each region to generate these comparisons.
This analysis determines the average distance between two cell types within a tissue region. We can then compare the average distance between different groups and samples to assess whether the proximity of two cell types varies across cohorts.