Entering edit mode

4 months ago

Rob
▴
160

I am using the following code to create a correlation plot between genes and a continuous variable. It gives me the correlation coefficient (R) and p-value on the plot. I want this to:

- give me FDR values
- How can I get R values and FDR values in a csv or excel file, Not just written on the image of plot?

code:

```
library(ggplot2)
library(plyr)
library(reshape2)
dat <- read.csv("Gene_variable_file.csv", check.name = FALSE, stringsAsFactors = FALSE, header = T)
df <- as.data.frame(dat)
# Basic scatter plot
ex <- melt(df, id.vars="continousVariable")
colnames(ex) <- c("continousVariable", "gene", "exprs")
#######################
####
ggscatter(ex, x = "continousVariable", y = "exprs",
add = "reg.line",
add.params = list(color = "blue", fill = "lightgray"),
color = "black", palette = "jco", fill = "lightgray",
#shape = "cyl",
fullrange = TRUE,
rug = TRUE, facet.by = "gene", cor.coef = T,
title = "Sarco-Specific DEGs-SMA Correlation-FEMALE",
conf.int = TRUE,
cor.coeff.args = list(),
cor.method = "spearman",
cor.coef.coord = c(NULL, NULL),
cor.coef.size = 4,
)+
geom_vline(xintercept = 102, colour="red", linetype = "longdash")
```

This is how my data look like:

Thank you. It tried

`broom`

from this link: https://cran.r-project.org/web/packages/broom/vignettes/broom.htmlBut which method I should use?

`ttest`

?`glm`

?`lm`

? I used lm and I got all my p-values 1,1,1,1... which cannot be correct.which one from the following code will be my data in broom code?"

`dat`

" or "`ex`

"I used

`ex`

as my data in`broom`

By the looks of things, it would be

`ex`

. You'd need to`group_by`

gene in some way. I've always used`group_by`

, but this vignette suggests`nest`

ing https://cran.r-project.org/web/packages/broom/vignettes/broom_and_dplyr.html.If you want to do pearson correlation, then

`lm`

should work fine, as the p-value on a linear model is identical to the p-value of a pearson's correlation. I'm not sure you can do spearman's that way though. You can use`cor.test(method="spearman")`

though.I agree that it is unlikely that all your p-values or 1. However, all your FDRs/adjusted p-values being 1 is entirely possible.