Our approach combines single cell RNA-seq (scRNA-seq) and bulk RNA-Seq data to relate the cellular state of one cell type (cell type A) to the cellular composition of the tumors (the abundance of cell type B). We applied it in melanoma to identify cell states of malignant cells which are associated with the "cold" phenotype, meaning, with low T cell infiltration levels.

You can use our approach to study T cell exclusion in other cancer types as well as the associations between other cell types within the tumor ecosystem. To do so, simply call

cell.cell.interactions.new.dataset(bulk.tpm,sc.tpm,sc.n.reads,
                                   cellA.markers,cellB.markers,
                                   cellA.name = "malignant.cell",
                                   cellB.name = "T.cell",
                                   bulk.confounders = NULL,
                                   sc.confounders = NULL,
                                   fileName = "Malignant.T.cell.exclusion",
                                   sigFilePath = "Malignant.T.cell.exclusion.signatures")

where

• bulk.tpm (G₁ x N₁) is a gene expression matrix of N₁ bulk tumors
• sc.tpm (G₂ x N₂) is a single-cell gene expression matrix of cell type A
• sc.n.reads (N₂ x 1) is a vector that corresponds to the cells in sc.tpm, and denotes the number of reads detected in each of the N₂ cells
• cellA.markers and cellB.markers are vectors of genes which are primarily expressed by cell type A and B, respectively
• cellA.name and cellB.name are the cell type names
• bulk.confounders (optional, N₁ x F₁) is a set of F₁ confounding factors in the N₁ bulk tumors, corresponding to bulk.tpm
• sc.confounders (optional, N₂ x F₂) is a set of F₂ confounding factors in the single-cell data, corresponding to sc.tpm
• fileName the name of the output rds file
• sigFilePath the name of the output csv file; this file will include the exclusion up (down) signature, consisting of genes which are up (down) regulated in cell type A in tumors with a low level of cell type B.