Hello,

I'm running MEDIPS.meth ,where the input(mset1 and iset1) are MEDIPS object obtained from MEDIPS.createSet. In the output table from running this function, everything is NA under the rpkm columns. My code and output warning are as below:

# Code for running MEDIPS.meth
> MEDIPS.meth(MSet1 = mset1, CSet = CS, ISet1 = iset1, CNV=FALSE,  minRowSum=10, diffnorm="tmm")

# Where the structure of mset object is:
> str(mset1)
Formal class 'MEDIPSset' [package "MEDIPS"] with 11 slots
  ..@ sample_name   : chr 
  ..@ path_name     : chr 
  ..@ genome_name   : chr "hg38"
  ..@ number_regions: int 195916427
  ..@ chr_names     : chr [1:22] "chr1" "chr2" "chr3" "chr4" ...
  ..@ chr_lengths   : num [1:22] 2.49e+08 2.42e+08 1.98e+08 1.90e+08 1.82e+08 ...
  ..@ genome_count  : int [1:11500016] 0 0 0 0 0 0 0 0 0 0 ...
  ..@ window_size   : int 250
  ..@ extend        : int 0
  ..@ shifted       : int 0
  ..@ uniq          : num 0.001

# Warning message
Warning messages:
1: In window_size * number_regions(MSet1[[i]]) :
  NAs produced by integer overflow
2: In window_size * number_regions(ISet1[[i]]) :
  NAs produced by integer overflow

Currently my workaround is adding a step to convert @number_regions and @window_size from type integer to numeric.

mset1@number_regions <- as.numeric(mset1@number_regions)
mset1@window_size <- as.numeric(mset1@window_size)

I'm wondering if the same thing happens on anyone else's end too when the number of regions is large? Thanks!

In the vignette, in section 4.2 under the diffnorm section it says you only get RPKM if you use diff.method = "ttest", and you chose the default, which is edgeR. I think this issue is orthogonal to the overflow warning.