strand-aware closest

[gfudenberg]

Example task:

Select CTCF peaks that are upstream of genes (in a strand-aware fashion). These peaks don't necessarily need to be on the same strand as the gene (hence would not be resolved just by adding the on=[] to closest()).

Solving this task currently requires multiple calls to closest.

Potential solutions:

• a convenience function that makes multiple calls to closest and merges the outputs.
• passing a column which defines the upstream/downstream relative to the genome. Default is relative to the reference genome (i.e. non-stranded closest).

Questions:

• if we pass this extra column, how to deal with a mix of stranded & non-stranded intervals? (where non-stranded is specified by '.': https://bioframe.readthedocs.io/en/latest/guide-specifications.html)
• what should this column be called? directionality_col, strand_col, stream_col

[gfudenberg]

is this issue closed @agalitsyna ?