CODEWITHSUNDEEP
r
Rnoob
Rnoob

Reputation: 1013

Select specific columns of a 'density' object in R

I need to pick out individual observations from a density object in R. This is related to a prior post except my data is formatted differently.

I begin with a matrix of gene expression values "b", where the columns represent 10 observations and rows represent expression levels for 1000 genes.

b = matrix(rexp(1000),ncol=10)

Next, I plot the densities with each observation as a separate line.

d = apply(b, 2, density, na.rm = TRUE)

plot(b, type="n", ylim=c(0,1))
for (i in 1:10) { lines( d[[i]] ) } 
abline(h=0.5)

How do I pick out the columns in 'b' whose densities are below the horizontal line at 'b[,1]=0.5'?

Upvotes: 0

Views: 141

Answers (1)

MrFlick
MrFlick

Reputation: 206606

If you want to get the estimated density at a particular value, you can use approx along with the points from the density object. For example

yv <- sapply(d, function(d) {
    approx(d$x, d$y, .5)[["y"]]
})
yv
# [1] 0.7158584 0.5516512 0.7049119 0.6180026 ...

This will predict the xvalue at x=.5 for each of the densities 10. To see which ones are below the horizontal line, you can do

which(yv<.5)
# [1] 5    # (your results may differ due to the random input)

So we see only the 5th column meets the criteria.

Upvotes: 1

Related Questions