Comparing different samples with Cytomap

Hi all, I have a large dataset of HE images. I have extracted cell features for each of them, clustered individual cells and now I’m trying to compare cluster composition for all samples. Basically I would like to generate a heatmap with x: clusters, y: all the samples. Is possible to do that with cytomap?
Thank you very much. Antonio

1 Like

Additional notes:
Cells were generated in QuPath.
Individual CSV files per sample.

I assume clusters were generated using all samples.

yes thank you Mike! we clustered individual cells using all samples. we can devide samples into 2 main molecular types. we would like to explore cluster composition in both types

Random thought: There are a lot of sample based heatmaps when you create neighborhoods. I am not sure about your exact use case, but you might try for neighborhoods of size 0, based on only your cluster measurement. That would get you access to all of the neighborhood analysis options, which have a lot of per-sample type of plots.

There is likely a better way to do this, but I do not know it off the top of my head.

There are a couple of things you can do here. The easiest is to take the clustered cells and use the Annotate Clusters function to pull out “phenotypes” for each cluster. This will save a new cell type for each cluster in each sample. You can name the clusters something useful if you know what each cluster is, in this example I left the default names.

Next you can use the Cullularity function to generate different types of heatmaps, bar graphs etc. that compare the number/percentage etc. of cells of each type across samples. In this interface you can set a group number for each sample type. This will put the heatmap in the order of groups. You can also use the size of the sample or image to convert cell numbers to density. Watch out for what you have selected in the divide by option because this can make your numbers look odd.

If you are comparing groups of samples you can use the Box plot by group number option to compare the number of cells from samples in each group for each cell type. As with most plots in CytoMAP the data from these can be exported into a .csv file if you want to use something like Prism to make the plots look nicer, or customize them.

Alternatively, you can use the region statistics function to look at the properties of the clustered cells. With this function you can make plots of the average channel intensity of each cluster (they are called regions in these plots but if you clustered cells these are really cell types). You can also make box plots of the number of cells of each type in each sample (this data can be exported to a .csv), or you can make stacked bar graphs of the % of each cell type for each sample.

1 Like

Most of these options should also work with cells clustered in CytoMAP.

1 Like

Ah, I have not once even tried to use the Cellularity button :slight_smile: Shows what I know!