populationRanges: multiple problems while I try to build on section "Identifying recurrent regions" of your vignette

[DaniilSarkisyan]

Dear Developers,

The functionality of populationRanges() looks very useful for our research, but I got the following problems while trying to build on section "Identifying recurrent regions" from your vignette. Can you help or suggest workarounds?

1. In addition to Gains/Losses we have CN-LOH type of CNVs. Technically it is "state=2", but it makes sense to include them as a third type of CNVs instead of just removing them. Do you have any suggestion how to do it?

2. Looking into your source code, populationRanges(..., est.recur = TRUE) calls .estimateRecurrence(pranges, grl) internally, which seems to give you a choice of mode = c("approx", "perm"). Unfortunately populationRanges() itself now has a parameter mode = c("density", "RO"), which makes methods "approx" and "perm" inaccessible.

3. It seems logical that populationRanges(..., mode = "RO", ro.thresh=0.1, classify.ranges=FALSE, est.recur = TRUE) would estimate recurrence without breaking CNVs into types. At least this is how the meaning of "classify.ranges=FALSE" is explained in your help.
   But it will try to use types during .estimateRecurrence() stage anyway and die with "Error in seq_len(len - 1) : argument must be coercible to non-negative integer" even on toy GRangesList example from populationRanges() help.

4. We would like to process ~15000 CNVs.
   But populationRanges(grl.IntOGen, mode = "RO", ro.thresh=0.1, classify.ranges=FALSE, est.recur = TRUE) keep dying with "Error in seq.default(1, length(both.scores) - 1, by = 2) : wrong sign in 'by' argument" even for the simplistic case of 3337 CNVs. We downloaded the list of genes from https://intogen.org/search, adding genomic regions using Bioconductor's Ensembl annotation and fabricated state=3 for ROLE=Act / state=1 for ROLE=LoF. I can upload you the 1.1M .rds file with GRangesList if you need reproducible example.

Thank you in advance,
Daniil Sarkisyan