Addition of a `Seq.join()` method

[widdowquinn]

Setup

I would like to be able to use the string-like concatenation/addition behaviour of Seq objects in the same way as using 'some_string'.join(), but this behaviour is not currently available.

Expected behaviour

spacer = Seq('NNNNN')
seqlist = list(SeqIO.parse('myfile.fasta', 'fasta'))
concatenated = spacer.join(seqlist)

Actual behaviour

Not implemented.

I appreciate that this is a minefield for resolving alphabets…

[peterjc]

If we leave aside mixing Seq and SeqRecord objects, I agree this would be nice and fits with the long term goal of making the Seq object more like the Pyhton string:

from Bio.Seq import Seq
from Bio import SeqIO

spacer = Seq('NNNNN')
concatenated = spacer.join(record.seq for record in SeqIO.parse('myfile.fasta', 'fasta'))

i.e. Define Seq object .join method taking an iterable of strings/other Seq-type objects.

This is relatively easy other than dealing with picking a consensus alphabet - although we can probably just follow the precedent elsewhere (e.g. for the MultipleSeqAlignment ),

https://github.com/biopython/biopython/blob/biopython-172/Bio/Alphabet/__init__.py#L324

In the special case of joining with an empty sequence, does its alphabet need consideration?

[widdowquinn]

Yeah, sorry about the mix-up of classes with SeqIO.parse() there… ;)

[kaskales]

@peterjc I see this issue is still open. Is it ok if I give it a shot as well?

[peterjc]

@kaskales did you have any thoughts on examples like Seq("").join(some_dna_seqs) vs Seq("", some_generic_alphabet).join(some_dna_seqs) where in some senses the default/generic alphabet of the empty sequence space is not important for the combined sequence (which in this example might be expected to be DNA from the alphabet attributes set in some_dna_seqs)

[kaskales]

@peterjc In your case where the spacer is an empty string, that would be the only time it might be good to assume the same alphabet as some_dna_seqs by default. Otherwise the spacer alphabet will take into consideration, where it should be up to the user to correctly identify the alphabet used in the spacer.

[peterjc]

Yes, I'm talking about the special case where the spacer is a zero length sequence.

[peterjc]

Applied as e856cd3 from #1746. Thank you all.