cell death events. In most cases, enzymatic treatment of the
tissue is required. Some examples of enzymes suitable for gen-
tle cell dissociation are Collagenase II (e.g., Roche Liberase
TM) or Papain (E.G. Worthington Biochemical Corporation).
Addition of calcium chelators such as EDTA can facilitate the
dissociation process and reduce the probability of
reaggregation.
Cell numbers and concentration can be quantified for
example using FACS, MACS or hemocytometer. Single-cell
mRNAseq methods typically require large numbers of input
cells, of which only a fraction is eventually captured.
In our case, we performed no labelling or staining of cells
prior to capture, allowing extremely short (<1 h) preparation
times before cell lysis, and obtained extremely high quality
scRNA data, while even DNA staining with Hoechst reduced
this quality significantly.6.The possibility of multiple cells existing per sample:Most of
the available scRNAseq methods (e.g., droplet- and FACS-
sorting based methods) capture cells blindly, i.e., it is not
possible to determine the number of cells per sample by imag-
ing of visual observation. Therefore, the possibility of multiple
cells per sample should be considered. Naturally, in the cases of
multiple cells in the sequencing sample the spatial mapping
should perform with lower confidence. Therefore, the cases
of multiple cells could be identified by lower mapping preci-
sion, or multiple mapped sites per sample, however, it is also
possible that the cell type of same molecular profile occurs in
multiple sites in the tissue or animal, complicating this inter-
pretation. Thus, the identification of cell numbers per sample
should include other parameters, such as number or transcripts
detected, the amount of cDNA, or ERCC spike-in ratios.
7.Considerations to the specificity score calculation: To calcu-
late specificity of each gene, we compare its level of expression in
a cell with its average expression level across all cells. This dam-
pens technical noise, and efficiently filters out housekeeping
genes that are expressed highly, but ubiquitously across all
cells. However, calculating the gene’s specificity only against
the background of sampled cells can lead to both false positive
and false negative calls, if the sampled cell set is small. We
recommend sampling at least 10x the number of cell types that
are hypothetically present in the tissue of interest, and even
higher numbers if some of the expexted cell types are under-
represented. Transcriptional bursts or random fluctuations in
expression can lead to an uncharacteristic, short-term high-
level expression of a gene, also resulting in a false-positive expres-
sion calls. In the mapping step, using a scoring function that
Spatial Transcriptomics 123