Trim a multiple sequence alignment based on a BED

As always, I assume there is a way to do this seqkit, but my fu is not strong enough to find it.

I have a BED file of positions/columns I would like to keep from a multiple sequence alignment.

I've tried seqkit subset -w 0 -U --bed keep.bed msa.fa > msa.trim.fa but I get the warning [WARN] sequence () not found in file: msa.fa and the resulting output is empty. I assume this is because the chromosome name in the BED does not match the names of the sequences in the MSA.

Esentially I want to ignore the chromosome name and just extract the positions in the BED from each sequence in the MSA.

Start by reviewing the seqkit subset command and its --bed handling, then compare that behavior with the reported MSA input and keep.bed coordinates. The work is done when BED positions can be applied to every sequence in the MSA without requiring the BED chromosome name to match a sequence name, and the resulting msa.trim.fa contains the requested columns.