bumphunter is failing build in Bioconductor 3.5 and 3.6 due to an update to the GenomicRanges and IRanges classes.
For example, the fourth test for 'Annotate Nearest' expects that the nearest Range to [4,5] will be [4,6] and not [2,3], so that the first few elements of annotateNearest(x,subject)$distance will be 0, 0, 0, 0, 0.
But the updated GenomicRanges and IRanges think that there are two equidistant Ranges to [4,5]. Both [4,6] and [2,3] are "zero-distance" from [4,5], so it makes an arbitrary choice and returns [2,3]. This makes the first few elements of annotateNearest(x,subject)$distance 0, -1, -1, -1, -1.
My inquiry to Bioconductor and their response is here:
bumphunter is failing build in Bioconductor 3.5 and 3.6 due to an update to the GenomicRanges and IRanges classes.
For example, the fourth test for 'Annotate Nearest' expects that the nearest Range to [4,5] will be [4,6] and not [2,3], so that the first few elements of annotateNearest(x,subject)$distance will be 0, 0, 0, 0, 0.
But the updated GenomicRanges and IRanges think that there are two equidistant Ranges to [4,5]. Both [4,6] and [2,3] are "zero-distance" from [4,5], so it makes an arbitrary choice and returns [2,3]. This makes the first few elements of annotateNearest(x,subject)$distance 0, -1, -1, -1, -1.
My inquiry to Bioconductor and their response is here:
https://support.bioconductor.org/p/100381/#100413