Results of TCGA data evaluation of candidate cell type markers

Cell type# Candidate genes# Selected markersMean pairwise similarity statistic in TCGASelected marker genes
B-cells3490.59BLK, CD19, FCRL2, MS4A1, KIAA0125, TNFRSF17, TCL1A, SPIB, PNOC
Cytotoxic cells18100.69PRF1, GZMA, GZMB, NKG7, GZMH, KLRK1, KLRB1, KLRD1, CTSW, GNLY
DC730.46CCL13, CD209, HSD11B1
Exhausted CD8540.44LAG3, CD244, EOMES, PTGER4
Macrophages3340.71CD68, CD84, CD163, MS4A4A
Mast cells3150.74TPSB2, TPSAB1, CPA3, MS4A2, HDC
Neutrophils3270.48FPR1, SIGLEC5, CSF3R, FCAR, FCGR3B, CEACAM3, S100A12
NK CD56dim cells1440.40KIR2DL3, KIR3DL1, KIR3DL2, IL21R
NK cells3630.47XCL1, XCL2, NCR1
T-cells1360.81CD6, CD3D, CD3E, SH2D1A, TRAT1, CD3G
Th1 cells271aNATBX21
CD8 T cells3520.51CD8A, CD8B
CD4 cells200bNA

aOnly one marker gene; quality impossible to assess in expression data alone

bCalculated as the T-cell score minus the CD8 cell score

Cell types lacking acceptable marker genes are omitted. The mean pairwise similarity statistic is a measurement of how well a gene set adheres to the co-expression patterns expected from a set of perfect marker genes, with a score of 1 indicating perfect marker-like behavior