Table 2.

Overview of the gold standard datasets

Dataset	Sequencing platform	Cell types	Number of samples	Mixture type	Cell quantification method
Cell mixtures	Illumina H12 Beadchip Microarray	B, mono, NK, neutrophil, CD4 T, CD8 T	12	In vitro cell mixtures	Controlled cell mixing in vitro
Simulated PBMC	Psuedo-bulk from scRNA-seq	B, mono, NK, CD4 T, CD8 T	200	In silico simulated mixtures	Simulated cell mixing in silico
Simulated stromal mixtures	Psuedo-bulk from scRNA-seq	B, CD4 T, CD8 T, endothelial, fibroblast, macrophages, mast cells	100	In silico simulated mixtures	Simulated cell mixing in silico
Framingham Cohort data	Affymetrix Human Exon ST 1.0 Microarray	Neutrophils, lymphocytes, monocytes	3728	Human clinical samples	Electrical impedance counting

Note: Each mixture comes from one of three platforms and consists of 12–3728 samples of varying sets of cell types. Mixtures were either created in silico, mixed in vitro via titration or taken directly from clinical patients. Underlying cell type fractions are either known naturally by nature of mixture generation or evaluated by electrical impedance counting.