How to pass external data to stat_compare_means in ggplot2 - r

Question

I created a barplot based on ggplot2 using an original long format that has been transferred with mean and standard error, which means I was not using raw data to generate a barplot with an errorbar but using processed mean and error data to generate a barplot. Therefore, I wonder how to add statistical comparison under such a situation using the original data because it is difficult to do statistics with only mean and error values.
This is my code, and you could imagine that df was the original long format data, and the data were grouped to calculate the mean and standard error, leading to the data summaryFGA.
The barplot with errorbar was generated using summaryFGA like below using the following code:
errorbar without statistical comparsion
summaryFGA <- df %>% group_by(DMP,tumor) %>% dplyr::summarize(mean = mean(FGA, na.rm = TRUE), se = std(FGA, na.rm = TRUE)) %>% as.data.frame
p1 <- ggplot(summaryFGA, aes(x = DMP, y = mean,fill = DMP)) +
geom_bar(stat = 'identity') +
geom_errorbar(aes(ymax = mean+se, ymin = mean-se),position = position_dodge(0.9), width = 0.15) +
scale_fill_manual(values = jco[1:2]) + scale_color_manual(values = jco[1:2]) +
ylab("Fraction of Genome Altered") + xlab("") +
facet_wrap(.~tumor, nrow = 1,scales = "fixed") +
# stat_compare_means(data = df,
# aes(x = DMP, y = FGA, fill=DMP),
# comparisons = my_comparisons,
# method = "wilcox.test", inherit.aes = F) +
theme(axis.text.x = element_text(hjust = 1, vjust = 0.5,size = 10, angle = 90,color = "black"),
axis.text.y = element_text(size = 10,color = "black"),
axis.ticks = element_line(size=0.2, color="black"),
axis.ticks.length = unit(0.2, "cm"),
panel.background = element_blank(),
panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
legend.position = "bottom",
strip.text = element_text(colour = 'white', size = 10),
axis.line = element_line(colour = "black"))
p1
g <- ggplot_gtable(ggplot_build(p1))
stripr <- which(grepl('strip-t', g$layout$name))
k <- 1
for (i in stripr) {
j <- which(grepl('rect', g$grobs[[i]]$grobs[[1]]$childrenOrder))
g$grobs[[i]]$grobs[[1]]$children[[j]]$gp$fill <- tumorColor[k]
k <- k+1
}
What I want is the following:
errorbar with statistical comparison
to add statistical comparison using the code I annotated, but it doesn't work. Obviously, it should be based on the original df data instead of summaryFGA.
If anyone wants an example, please refer to this thread and add a statistical comparison using "original data" in the example.
Many thanks in advance.