plot overlapping positions from different files in R

Question

I have a two big files with Start and End positions and two sample columns (numeric).

File 1:

Start  End  Sample1  Sample2
1         60      1       4
100       200     2       1
201       250     1       4
300       450     1       1

File 2:

Start  End  Sample1  Sample2
40         60      1       1
70        180      1       1
240       330      2       1
340       450      1       4
500       900      1       4
980       1200     2       1

First, I would like to take the first Start and End positions from first file and make a segment plot. The plot must also take into account Start-20 and End+20 for each position in first file.

Then I would like to take the overlapping Start and End positions from the second file and plot it on the above plot. In this way there will be many plots based on the Start and End positions from the first file and the one with out overlaps will also be plotted individually.

The color for each segment will be based on the two sample numbers (for example in both the files if its 1 and 4 the color of segment will be red, if its 1 and 1 the color of segment will be green and so on).

I would really appreciate if someone make me understand how to make function for this in R.

Thanks in advance.

PS I have attached the drawing for an output. I have shown only two results.

Below is the code which I wrote but it gives an error

Error in match.names(clabs, names(xi)) : names do not match previous names

Also I need to specify red color to dataset1 line segment and green color to dataset2 line segments. How will I implement it in the code below?

overlap_func <- function(dataset1,dataset2) {

for(i in 1:nrow(dataset1))
 {

 loop_start <- dataset1[i,"Start"]
 loop_end <- dataset1[i,"End"]
 p <- dataset2[,c(1,2)]   
 dataset1_pos <- data.frame(loop_start,loop_end)
 dataset2_filter <- p[p$Start >= (loop_start-(loop_start/2)) & p$End <= (loop_end+ (loop_end/2)), ]
 data_in_loop <- rbind(dataset1_pos,dataset2_filter)
 plot_function(data_in_loop,loop_start,loop_end)

 }
 }


plot_function <- function(loop_data,start,end){ 
 pos <- 1:nrow(loop_data)
 dat1 <- cbind(pos,loop_data)
 colnames(dat1) <- c("pos","start","end")
 pdf(file=paste0("path where plots are generated","_",start,"-",end,"_","overlap.pdf"))
 plot(dat1$pos, type = 'n', xlim = range(c(start-(start/2), end+(end/2))))
 segments(dat1$start, dat1$pos, dat1$end, dat1$pos)
 dev.off()
 }


df1 <- read.table(header=T, text="Start  End  Sample1  Sample2
1         60      1       4
100       200     2       1
201       250     1       4
300       450     1       1")

df2 <- read.table(header=T, text="Start  End  Sample1  Sample2
40         60      1       1
70        180      1       1
240       330      2       1
340       450      1       4
500       900      1       4
980       1200     2       1")

 overlap_func(df1,df2)

Answer 1

Something like this??

df1 <- read.table(header=T, text="Start  End  Sample1  Sample2
1         60      1       4
100       200     2       1
201       250     1       4
300       450     1       1")

df2 <- read.table(header=T, text="Start  End  Sample1  Sample2
40         60      1       1
70        180      1       1
240       330      2       1
340       450      1       4
500       900      1       4
980       1200     2       1")

require(IRanges)
require(ggplot2)
require(plyr)

df1$id <- factor(1:nrow(df1))
ir2 <- IRanges(df2$Start, df2$End)
out <- ddply(df1, .(id), function(x) {
    ir1 <- IRanges(x$Start, x$End)
    o.idx <- as.data.frame(findOverlaps(ir1, ir2))$subjectHits
    df.out <- rbind(x[, 1:4], df2[o.idx, ])
    df.out$id1 <- x$id
    df.out$id2 <- seq_len(nrow(df.out))
    df.out
})
out$id1 <- factor(out$id1)
out$id2 <- factor(out$id2)
out$id3 <- factor(1:nrow(out))

p <- ggplot(out, aes(x = Start, y = id3 , colour = id2)) 
p <- p + geom_segment(aes(xend = End, ystart = id3, yend = id3))
p <- p + scale_colour_brewer(palette = "Set1")
p

Edit: After looking at your updated drawing, maybe this is what you want?

p + facet_wrap( ~ id1, scales="free")

Edit: Save each plot in facet in separate file. You can do this by generating the plot each time by splitting on id1

d_ply(out, .(id1), function(ww) {
    p <- ggplot(ww, aes(x = Start, y = id3 , colour = id2)) 
    p <- p + geom_segment(aes(xend = End, ystart = id3, yend = id3))
    p <- p + scale_colour_brewer(palette = "Set1")
    fn <- paste0("~/Downloads/id", as.numeric(as.character(ww$id1[1])), ".pdf")
    ggsave(fn, p)
})

Set the path in fn accordingly.

Answer 2

I tried to solve this problem using lattice package. I specially use function Shingle to get an idea of the intervals to compare. I hoped I can merge the 2 shingles but I can't. So, once I have a my first plot, I used ( as in the above solution) IRanges package to compute the overlaps. The idea is have a final dotplot.

## I red the input data
dat <- read.table(text = 'Start  End  Sample1  Sample2
1         60      1       4
100       200     2       1
201       250     1       4
300       450     1       1', header = T) 

dat1 <- read.table(text = 'Start  End  Sample1  Sample2
40         60      1       1
70        180      1       1
240       330      2       1
340       450      1       4
500       900      1       4
980       1200     2       1', header = T) 


## I create my 2 shingles
dat.sh <- shingle(x = dat[,3], intervals = dat[,c(1,2)])
dat1.sh <- shingle(x = dat1[,3], intervals = dat1[,c(1,2)])
## compute max value for plot comparison
max.value <- max(c(dat$End,dat1$End))
## I plot the 2 series with differents color
p1<- plot(dat.sh, xlim= c(0,max.value),col = 'red')
p2 <- plot(dat1.sh,xlim= c(0,max.value), col ='green')
library(gridExtra)
grid.arrange(p1,p2)

This is a fast method to compare my intervals.

This looks Ok, but I can't go further with shingle since I can't merge them in the same plot. SO I will use IRanges package to compute the overlaps.

library(IRanges)
rang1 <- IRanges(start=dat[,1], end = dat[,2])
rang2 <- IRanges(start=dat1[,1], end = dat1[,2])
dat.plot     <- dat1                    # use the first data.frame
dat.plot$group <- 'origin'
dat.plot$id <- rownames(dat1)           ## add an Id for each row
rang.o <- findOverlaps(rang2,rang1)     # get overlaps
dat.o <- dat1[rang.o@queryHits,]        ## construct overlaps data.frame 
dat.o$id <- rang.o@subjectHits
dat.o$group <- 'overlap'
dat.plot <- rbind(dat.plot,dat.o)      ## union of all 
dotplot(id ~End-Start|group  , data=dat.plot, 
                               groups = col,type = c("p", "h"))