Extended Data Fig. 2: TACCO splitting of mixed expression profiles into pure constituents. | Nature Biotechnology

Extended Data Fig. 2: TACCO splitting of mixed expression profiles into pure constituents.

From: TACCO unifies annotation transfer and decomposition of cell identities for single-cell and spatial omics

Extended Data Fig. 2: TACCO splitting of mixed expression profiles into pure constituents.

UMAP embeddings of in-silico mixed mouse colon scRNA-seq profiles (top four rows: real or balanced type composition and with and without downscaling the count-yield per cell by a factor of 10) or Slide-seq beads (bottom row), based on ground truth (far left, where available), TACCO annotation (second left), reference data (second right) and mixed expression data split using TACCO’s annotation as input embedded in a joint UMAP of the concatenated reference-split dataset (rightmost). The in silico mix uses a beadsize of 1.0. For the joint embedding data was filtered to include only observations with at least 30 counts in genes which were annotated highly variable in the reference dataset, except for capture_rate = 1.0 where a threshold of 100 is used.

Back to article page