结合数据中的图表并创建图表列表 R

Question

I think this may be a easy fix but I cannot get it to work. This is for a R-shiny app down the line which has two input data sets structured like this:

d1 <- data.frame(A1 = rnorm(10),
             A2 = rnorm(10),
             A3 = rnorm(10),
             B1 = rnorm(10),
             B2 = rnorm(10),
             B3 = rnorm(10),
             C1 = rnorm(10),
             C2 = rnorm(10),
             C3 = rnorm(10),
             Names = c("A", "B", "C", "D", "E",
                       "F", "G", "H", "I", "J"))

While this data is structured in a logical way representing replicates of the same sample this is not guaranteed in all cases and inputs vary from user to user. Because of this I prompt for an index to indicate which columns are related (ie they get the same name) which gives a data frame that looks like this:

d2 <- data.frame(ID = c("A1", "A2", "A3", 
                    "B1", "B2", "B3",
                    "C1", "C2", "C3"),
             Anno = c(rep("A", 3), 
                      rep("B", 3),
                      rep("C", 3)), stringsAsFactors = FALSE)

I can compare each sample to every other sample with a scatterplot using this method

plot_combinations = combn(names(d1)[1:ncol(d1)-1], 
                      2, 
                      simplify = FALSE)
plot_list <- list()
plot.col = "black"
for (i in 1:length(plot_combinations)) {
  p = ggplot(d1, 
             aes_string(x = plot_combinations[[i]][1], 
                        y = plot_combinations[[i]][2])) +
    geom_point(pch = 21, colour = "black", fill = plot.col) +
    theme_classic(base_size = 14)
  plot_list[[i]] = p
}

While in principle this works it is not feasible as with a small data set this is already 36 combinations. What makes more sense is to compare within replicates instead. In order to do that I can rename the columns of d1 and get the frequency of each replicate and save it in a variable sampleFreq and extract data by looping over the replicate names like this:

anno <- d2$Anno
d3 <- d1[,1:nrow(d1)-1]
colnames(d3) <- anno
sampleFreq <- as.data.frame(table(d2[,2]))
datalist <- list()
for (i in sampleFreq[, 1]) {
  d4 <- d3[, i == names(d3)]
  datalist[[i]] = d4
}

This gives me a list of data frames structured by replicate and I would like to create a list of scatter plots from this. I have tried for loops but run into issues, I think there may be a lapply solution to this but I am stuck. The idea is to then call the plots from the plot list using a index number determined by number of times a button is clicked in shiny.

Any help would be awesome

Thanks!

Answer 1

What not do a pair plot within each replicate group? We can start with your data, I try not to rename the columns in d1 to d3(like you did)

d1 <- data.frame(A1 = rnorm(10),
             A2 = rnorm(10),
             A3 = rnorm(10),
             B1 = rnorm(10),
             B2 = rnorm(10),
             B3 = rnorm(10),
             C1 = rnorm(10),
             C2 = rnorm(10),
             C3 = rnorm(10),
             Names = c("A", "B", "C", "D", "E",
                       "F", "G", "H", "I", "J"))

d2 <- data.frame(ID = c("A1", "A2", "A3", 
                    "B1", "B2", "B3",
                    "C1", "C2", "C3"),
             Anno = c(rep("A", 3), 
                      rep("B", 3),
                      rep("C", 3)), stringsAsFactors = FALSE)

We go through all replicates:

pdf(<somefilename.pdf>)
for(i in unique(d2$Anno)){
COLS=d2$ID[d2$Anno ==i]
pairs(d1[,COLS],upper.panel=NULL)
}
dev.off()

Looks like this but it's (bloody) fast

Or you can use GGally,

library(GGally)

plots <- lapply(unique(d2$Anno),function(i){

    COLS=as.character(d2$ID[d2$Anno ==i])
    return(ggscatmat(d1[,d2$Anno %in% i]))
    })

plots[[1]]