Extract genomic feature related to point from mean-variance plot in limma/voom

Question

[James Ashmore]

I am using limma/voom for differential expression analysis and I would like to find out which genomic feature is represented by each point in the mean-variance modelling plot which voom generates. There are some weird trends happening in the data and I'd like to see which genomic features these correspond to. I tried using the save.plot option in the call to voom but it doesn't produce the voom.xy value in the output object?

Answer 1

[h.mon]

Old thread, I know, but Biostar bumped it.

With:

v <- voom( x, design, plot = T, save.plot = T )
identify( v$voom.xy, labels = row.names( v ) )

I could click on points and, upon pressing esc, the names of the points clicked were printed on the graph, and their index at row.names( v ) returned. You can also assign the indexes to a variable:

z <- identify( v$voom.xy, labels = row.names( v ) )
z
#[1]  634 3097 4041 9483

rownames(v)[z]
#[1] "C16H1orf115" "POC1A"       "PLPP2"       "FADD"

You also (probably) can accomplish this with the Glimma package, which I never used but seems really great.