log 0 and multiplicity of the p-value

[antpiron]

Hi,

I have observed that, on large list, the p-value underflow becomes exactly zero. In this case, the package takes log 0 and it results into an infinite result.
Here is a code to reproduce the issue:

library(RRHO)

df <- data.frame(gene = 1:5000, a = 1:5000, b = 1:5000)

RRHO_obj <-  RRHO::RRHO(df[, c("gene", "a")], df[, c("gene", "b")],
                        labels = c("a", "b"), alternative = "two.sided", plots = TRUE,
                        outputdir = "/tmp/")

The overlap map RRHOMapa_VS_b.jpg shows a big white patch for infinite. I think that it would be reasonable to replace the log 0, with the log of 4.94065645841247E-324 (see the smallest number above 0 for doubles, https://learn.microsoft.com/en-us/dotnet/api/system.double.epsilon?view=net-7.0 ).

Another issue is that sometime the minimal p-value is not unique, in this case the coordinate returned by

maxind.lr  <- which(
    max(hypermat.signed[1:(ceiling(nrow(hypermat.signed)/2)-1), 
                   1:(ceiling(ncol(hypermat.signed)/2)-1)],
        na.rm=TRUE) == hypermat.signed, arr.ind=TRUE)

is wrong because maxind.lr is then a matrix (not a vector) and maxind.lr[2] is the x of the second coordinate (and not the y of the first).

Thank you for your package.