add pvalues to final spillover-reduction figure

notebooks/70_pub_figures.Rmd

@@ -413,34 +413,39 @@ migration_legend = cowplot::get_legend(p)
 
 ```{r, include=FALSE}
 # Plot differences with removed marker genes
-rmgenes_plot_dc = res_spillover$rm_marker_genes_dc %>%
+rmgenes_data_dc = res_spillover$rm_marker_genes_dc %>%
   inner_join(method_names) %>%
   filter(cell_type == "B cell") %>%
-  mutate(dataset = if_else(dataset == "before", "no", "yes")) %>%
+  mutate(dataset = if_else(dataset == "before", "no", "yes"))
+
+rmgenes_plot_dc = rmgenes_data_dc %>%
   ggplot(aes(x=dataset, y=predicted_fraction, colour=dataset)) +
     geom_quasirandom() +
     stat_summary(fun.y=mean, geom="crossbar", fun.ymin=mean, fun.ymax=mean, width=.5, color="black") +
     facet_wrap(~method_name, drop = TRUE) +
-    stat_compare_means(paired = TRUE, method = "t.test", label="p.signif") +
+    stat_compare_means(paired = TRUE, method = "t.test", label="p") +
     theme(legend.position = "none") +
     xlab("signature genes removed") +
     ylab("predicted B cell score") +
     scale_color_brewer(type="qual", palette=6, direction = -1) +
     ylim(0, .75) +
     theme(plot.margin = margin(t=1, unit = "cm"))
 
+
 ```
 
 ```{r, include=FALSE}
-rmgenes_plot_cd4 = res_spillover$rm_marker_genes_cd4 %>%
+rmgenes_data_cd4 = res_spillover$rm_marker_genes_cd4 %>%
   inner_join(method_names) %>%
   filter(cell_type == "T cell CD4+") %>%
-  mutate(dataset = if_else(dataset == "before", "no", "yes")) %>%
+  mutate(dataset = if_else(dataset == "before", "no", "yes"))
+
+rmgenes_plot_cd4 = rmgenes_data_cd4 %>% 
   ggplot(aes(x=dataset, y=predicted_fraction, colour=dataset)) +
     geom_quasirandom() +
     stat_summary(fun.y=mean, geom="crossbar", fun.ymin=mean, fun.ymax=mean, width=.5, color="black") +
     facet_wrap(~method_name, drop = TRUE) +
-    stat_compare_means(paired = TRUE, method = "t.test", label="p.signif") +
+    stat_compare_means(paired = TRUE, method = "t.test", label="p") +
     theme(legend.position = "none") +
     xlab("signature genes removed") +
     ylab("predicted CD4⁺ T cell score") +
@@ -451,6 +456,19 @@ rmgenes_plot_cd4 = res_spillover$rm_marker_genes_cd4 %>%
 
 ```
 
+```{r spillover_correction_pvalues, include=FALSE}
+tmp_datas = list("cd4_background"=rmgenes_data_cd4, "b_dc_spillover"=rmgenes_data_dc)
+lapply(names(tmp_datas), function(which_data) {
+  lapply(c("epic", "quantiseq", "cibersort"), function(method) {
+    tmp_data = tmp_datas[[which_data]] %>% filter(method == !!method) %>% 
+      spread(dataset, predicted_fraction)
+    cat(sprintf("## Testing before/after filtering unspecific genes for %s for method %s", which_data, method))
+    print(t.test(tmp_data$yes, tmp_data$no, paired = TRUE, alternative = "two.sided"))
+    cat(sprintf("-> Reduction by %f", (mean(tmp_data$no)-mean(tmp_data$yes))/mean(tmp_data$no) ), "\n\n")
+  })
+})
+```
+
 ```{r fig.width = 8, fig.height=9, echo=FALSE, message=FALSE, fig.cap="Migration chart figure for paper. "}
 # combine legend and plot
 plot_lower = plot_grid(rmgenes_plot_cd4, rmgenes_plot_dc,