raw counts plus metadata such that one can do a second analysis where one keeps only cells in cluster 1 (this could be done via the filter options)
counts/data/z-scores for a representative set of cells (e.g. 500) as table together with metadata so that the user can plot their own heatmaps as csv/excel
Let's collect ideas of what else would be useful to add to the output: