Hello,

I find it odd that coersion to GRanges with as fails in the case of an empty data.frame:

This works fine:

> as(data.frame(SeqID=c('s'),start=c(1),end=c(2)),'GRanges')
GRanges object with 1 range and 0 metadata columns:
      seqnames    ranges strand
         <Rle> <IRanges>  <Rle>
  [1]        s    [1, 2]      *
  -------
  seqinfo: 1 sequence from an unspecified genome; no seqlengths

But if the data.frame is empty and untyped, no dice:

> as(data.frame(SeqID=c(),start=c(),end=c()),'GRanges')
Error in .find_start_end_cols(df_colnames0, start.field0, end.field0) : 
  cannnot determine start/end columns

Taking the diagnostic literally, I type the columns in the data.frame and get a little further

as(data.frame(SeqID=character(),start=integer(),end=integer()),'GRanges')
Error in validObject(.Object) : 
  invalid class "GRanges" object: NROW(strand(x)) != length(x)

Almost across the line. Let's add the strand in too:

> as(data.frame(SeqID=character(),start=integer(),end=integer(),strand=character()),'GRanges')
GRanges object with 0 ranges and 0 metadata columns:
   seqnames    ranges strand
      <Rle> <IRanges>  <Rle>
  -------
  seqinfo: no sequences

That works.

But, is this truely by design?

At the very least, I would hope that the 3rd example would have worked, with the strand column optional...

Or is there a "good reason" it works just as it does?

Thanks!