Here we will be calculating radial distance from tumor regions using semla (Larsson et al. 2023) R package, then fit a spline to observe the trend of Fibro2 and Fibro3 scores (and their genes) as a function of distance from the tumor regions.
xenium_samples <- qread("outputs/xenium_obj_harmony_integrated.qs")
xenium_samples <- SplitObject(xenium_samples, split.by = 'sample_id')
xenium_samples <- lapply(xenium_samples, function(x){
x <- UpdateSeuratForSemla(x)
x <- RadialDistance(
x,
column_name = "annotation",
selected_groups = c("Tumor"),
maxDist = 100
)
x$r_dist_Tumor_sqrt <- sign(x$r_dist_Tumor)*sqrt(abs(x$r_dist_Tumor))
return(x)
})fiorini_data[[]] |>
filter(annotation == 'Fibroblasts' & r_dist_Tumor_sqrt > 0) |>
select(fibro2_score_scaled, fibro3_score_scaled, r_dist_Tumor_sqrt) |>
pivot_longer(c("fibro2_score_scaled", "fibro3_score_scaled"), names_to = "Signature", values_to = "Score") |>
ggplot(aes(r_dist_Tumor_sqrt, Score, color = Signature)) +
geom_smooth(method = "gam", formula = y ~ s(x, bs = "cs"), linewidth = 2) +
geom_vline(aes(xintercept = 0, color = "border"), linetype = "dashed") +
xlab("sqrt(Radial Distance from Tumor)") + ylab("Fibroblast-specific Fibro2/3 Score") +
scale_color_manual(values = fibro_sig_cols) +
theme_bw()fibro_markers <- read.csv("../Fibro Top Markers.csv", row.names = 1)
fibro2tumor_vs_fibro3tumor <- read.csv("~/marina_turbo/Visium_Project/analysis/outputs/Fibroblasts_Analysis/stroma2_tumor_v_stroma3_Tumor.csv")
fibro2tumor_vs_fibro3tumor %>%
filter(padj < 0.05, log2FoldChange > 0.5) %>%
pull(X) %>%
intersect(rownames(xenium_samples)) -> fibro2_markers
fibro3tumor_vs_fibro3gol <- read.csv("~/marina_turbo/Visium_Project/analysis/outputs/Fibroblasts_Analysis/stroma3_Tumor_v_stroma3_GoL.csv")
fibro3tumor_vs_fibro3gol %>%
filter(padj < 0.05, log2FoldChange > 0.5) %>%
pull(X) %>%
intersect(rownames(xenium_samples)) -> fibro3_tumor_markers
fibro3tumor_vs_fibro3gol %>%
filter(padj < 0.05, log2FoldChange < -0.5) %>%
pull(X) %>%
intersect(rownames(xenium_samples)) -> fibro3_gol_markers
fibro2_genes <- c("ACTA2", "TAGLN", "INHBA", "LOXL2", "SFRP4", "WNT5A", "MFAP5")
data <- xenium_samples[[]] |>
bind_cols(FetchData(xenium_samples, vars = fibro2_genes)) |>
filter(annotation == 'Fibro') |>
pivot_longer(all_of(fibro2_genes), names_to = "Gene", values_to = "Expression") |>
select(sample_id, r_dist_Tumor, Gene, Expression)
data$sample_id <- factor(data$sample_id)
fibro2_genes <- c("ACTA2", "TAGLN", "INHBA", "LOXL2", "SFRP4", "WNT5A", "MFAP5", "LRRC15")
xenium_samples[[]] |>
bind_cols(FetchData(xenium_samples, vars = fibro2_genes)) |>
filter(annotation == 'Fibro') |>
pivot_longer(all_of(fibro2_genes), names_to = "Gene", values_to = "Expression") |>
ggplot(aes(r_dist_Tumor, Expression, color = Gene)) +
geom_smooth(method = "gam", formula = y ~ s(x, bs = "cs")) +
geom_vline(aes(xintercept = 0, color = "border"), linetype = "dashed") +
xlab("Radial Distance from Tumor Epithelial") + ylab("Fibroblasts-specific Expression")
fibro3_genes <- c("C7", "EDNRB", "IGF1")
xenium_samples[[]] |>
bind_cols(FetchData(xenium_samples, vars = fibro3_genes)) |>
filter(predicted_cellType == 'Fibroblasts') |>
pivot_longer(all_of(fibro3_genes), names_to = "Gene", values_to = "Expression") |>
ggplot(aes(r_dist_Tumor_Epithelial, Expression, color = Gene)) +
geom_smooth(method = "gam", formula = y ~ s(x, bs = "cs")) +
geom_vline(aes(xintercept = 0, color = "border"), linetype = "dashed") +
xlab("Radial Distance from Tumor Epithelial") + ylab("Fibroblasts-specific Expression")