Explore the cell-by-cell data

n=1, FFPE, exploratory.

Two small biopsies from a single patient, formalin-fixed and paraffin-embedded. It generates hypotheses that other tests will have to confirm.

RNA ≠ protein; IHC decides.

The analysis reads RNA. Reading a gene does not mean the protein is there; immunohistochemistry tells you that.

Ambient floor.

In a sample like this, loose RNA floats around and lands in any cell. For each gene and group we compute how many cells would come out positive from that alone; anything that does not clear the floor does not count. In the liver it is a minimum: for tumor genes it falls short, and the matched reference is what decides. Also, probes that fall inside an exon read genomic DNA from the tissue too; it weighs more in the bone, and its floor includes it. Next to each gene you will see how many of its probes span an exon junction.

Standard Cell Ranger matrix, all probes; in the bone it includes genomic DNA.

These are the counts Cell Ranger gives with every probe on the panel, the same basis as the reviewed analysis.

Circular figures.

If a gene was used to choose a group’s cells, its figure in that group comes out high (or low) by construction. Those genes carry a label and, where the review measured them without that circularity, that other figure next to it.

Median per detected cell; the review gives pseudobulk, so level figures need not match.

The level the page draws is the median in cells where the gene is detected; the review gives the level of the whole population pooled.

Polaris reading.

The per-gene reading is Polaris’s and is a draft, pending the laboratory’s human reading. It does not speak for the laboratory.

Matched reference.

Non-tumor cells from the same sample, at the same depth (molecules per cell); in the liver they are T and NK cells, myeloid cells and endothelium; hepatocytes and fibroblasts are not included. If a gene is detected as much or more in the reference than in the group of interest, it does not tell the tumor apart.

Fixed groups.

The groups and their n come from the reviewed analysis. The page does not regroup cells or compute a “% of tumor cells” for the liver.

The text summary of these data is in Cell by cell.