从edgeR中的BCV图中查找坐标

Question

我如何知道函数（x），从而知道从edgeR包生成色散图（BCV）的函数的坐标？

我需要知道＆＃34;坐标＆＃34; ＆＃34;趋势＆＃34;的顶点（每个顶点的最大值和最小值）分散和＆＃34;趋势线（蓝色）＆＃34;与公共色散线相交（红线）

为方便起见，我做了一个例子：

data <- "chr start end depth depth1 depth2 depth3 depth4 depth5 depth6
chr1 3273 3273 7 200 35 1 200 850 0
chr1 3274 3274 3 50 25 5 300 1500 2
chr1 3275 3275 8 600 15 8 100 300 5
chr1 3276 3276 4 30 2 10 59 20 0
chr1 3277 3277 25 20 7 4 600 45 0"
data <- read.table(text=data, header=T)

datamatrix <- data [, c(4:10)]
library (edgeR)

#grouping factor
group <- c(1, 2, 2, 2, 2, 2, 2) #groups of each sample)

#create a DGEList
y <- DGEList(counts=datamatrix,group=group)

根据edgeR包，我可以使用以下方法估算数据集的分散度：

y <- estimateDisp(y) 

y$common.dispersion

共同色散的平方根给出了生物变异系数。并且，可以在BCV图中查看色散估计值：

plotBCV(y)

Answer 1

我认为这解决了这个伎俩

#finding dispersion coordinates

    afun <- approxfun(y$AveLogCPM, y$trended.dispersion)

#ranging between -2 and 12 the "par" number to check the intersection
    optim(par=-1, function(x) { (afun(x) - y$common.dispersion)^2 }) #-0.9056
    optim(par=0, function(x) { (afun(x) - y$common.dispersion)^2 }) #-0.4955
    optim(par=2, function(x) { (afun(x) - y$common.dispersion)^2 }) #1.8237

#with the plotBCV function code, I have the access to tagwise, trended, common dispersions, and the abundance values, for each CpG as: y$tagwise.dispersion, y$trended.dispersion, y$common.dispersion, and y$AveLogCPM, respectively. 

min (y$trended.dispersion) #[1] 1.09009
max (y$trended.dispersion) #[1] 1.50691

#to the vertex, you can subset a table according to the AveLogCPM range that you want to know the max and min like this:

afuntable<- data.table(y$AveLogCPM, y$trended.dispersion)
afuntablemod <-subset (afuntable, V1 >= 0)
afuntablemod <-subset (afuntablemod, V1 <= 2)
min (afuntablemod$V2) #[1] 1.318002

#and again
afuntable<- data.table(y$AveLogCPM, y$trended.dispersion)
afuntablemod <-subset (afuntable, V1 >= 2)
afuntablemod <-subset (afuntablemod, V1 <= 12)
max (afuntablemod$V2) #[1] 1.50691