distance_to_locus is inf for loci on opposite strands, feature or a bug?

[rraadd88]

Locus object's distance_to_locus attribute calculates the distance between the loci on opposite strands, as inf.

Demo code:

from pyensembl.locus import Locus
## loci
locus1=Locus(contig="1", start=1, end=2,strand='+')
locus2=Locus(contig="1", start=4, end=5,strand='-')
## distance
print(locus1.distance_to_locus(locus2))
## change the strands to be the same
locus2.strand='+'
## distance
print(locus1.distance_to_locus(locus2))

### output
# inf
# 2

Because unifying the strands gives the correct distance, I wonder if this is a bug, or if it is supposed to be a feature for e.g. maybe to discourage calculating distances between loci on opposite strands.

[iskandr]

This was intentional since "distance" on opposite strands wasn't useful for my usecases when I started working on PyEnsembl. Since making it strand-invariant changes the semantics, would you be OK with a different helper function?

I'm not sure what to call it distance_to_locus_on_either_strand, strand_invariant_distance_to_locus, ... ?

[rraadd88]

Ok, thank you for your answer.

As for the name of the new helper function, to make it easy to find, I would personally like if it would have the distance_to_locus prefix, as in distance_to_locus_on_either_strand. That way, this variant function could be easily found once one knows the name of the original function.

Adding a new helper function would do it. But how about adding a parameter, called ignore_strands for example, to the existing function instead? Could that be easier/simpler? Asking because when I came across this issue, intuitively, the first thing I did was to look for the parameter that could allow me to calculate the distance in a strand-invariant way.