5. Draw cell-cell colocalization heatmap
# Load global overlap score calculated between cancer epithelial and other cell types
df <- sp_data_grid_celltype@misc %>%
dplyr::select(Feat_1, Feat_2, J_comp) %>%
tidyr::pivot_wider(names_from = Feat_1, values_from = J_comp) %>%
tibble::column_to_rownames(var = "Feat_2")
# Fill NA values with 0
df[is.na(df)] = 0
# Color choices
grDevices::hcl.pals()
#> [1] "Pastel 1" "Dark 2" "Dark 3" "Set 2"
#> [5] "Set 3" "Warm" "Cold" "Harmonic"
#> [9] "Dynamic" "Grays" "Light Grays" "Blues 2"
#> [13] "Blues 3" "Purples 2" "Purples 3" "Reds 2"
#> [17] "Reds 3" "Greens 2" "Greens 3" "Oslo"
#> [21] "Purple-Blue" "Red-Purple" "Red-Blue" "Purple-Orange"
#> [25] "Purple-Yellow" "Blue-Yellow" "Green-Yellow" "Red-Yellow"
#> [29] "Heat" "Heat 2" "Terrain" "Terrain 2"
#> [33] "Viridis" "Plasma" "Inferno" "Rocket"
#> [37] "Mako" "Dark Mint" "Mint" "BluGrn"
#> [41] "Teal" "TealGrn" "Emrld" "BluYl"
#> [45] "ag_GrnYl" "Peach" "PinkYl" "Burg"
#> [49] "BurgYl" "RedOr" "OrYel" "Purp"
#> [53] "PurpOr" "Sunset" "Magenta" "SunsetDark"
#> [57] "ag_Sunset" "BrwnYl" "YlOrRd" "YlOrBr"
#> [61] "OrRd" "Oranges" "YlGn" "YlGnBu"
#> [65] "Reds" "RdPu" "PuRd" "Purples"
#> [69] "PuBuGn" "PuBu" "Greens" "BuGn"
#> [73] "GnBu" "BuPu" "Blues" "Lajolla"
#> [77] "Turku" "Hawaii" "Batlow" "Blue-Red"
#> [81] "Blue-Red 2" "Blue-Red 3" "Red-Green" "Purple-Green"
#> [85] "Purple-Brown" "Green-Brown" "Blue-Yellow 2" "Blue-Yellow 3"
#> [89] "Green-Orange" "Cyan-Magenta" "Tropic" "Broc"
#> [93] "Cork" "Vik" "Berlin" "Lisbon"
#> [97] "Tofino" "ArmyRose" "Earth" "Fall"
#> [101] "Geyser" "TealRose" "Temps" "PuOr"
#> [105] "RdBu" "RdGy" "PiYG" "PRGn"
#> [109] "BrBG" "RdYlBu" "RdYlGn" "Spectral"
#> [113] "Zissou 1" "Cividis" "Roma"# Draw heatmap with pheatmap
out <- pheatmap::pheatmap(t(df), cluster_rows=T, show_rownames=T,
cluster_cols=T, angle_col=45, fontsize=5,
color = grDevices::hcl.colors(100, "Inferno"),
na_col = "grey70",
breaks = seq(0, 0.3, by = 0.01*0.3),
clustering_method = 'average',
clustering_distance_rows = "correlation",
clustering_distance_cols = "correlation")
