Warning in GFFParser: BiopythonDeprecationWarning: UnknownSeq(length) is deprecated; please use Seq(None, length) instead.

[victorlin]

Scope

bcbio-gff 0.7.1 and Biopython 1.79 or 1.80.

Description

The feature in question was deprecated in Biopython 1.79 and removed in 1.81. While GFFParser has been written to handle the removal, it had silently ignored the deprecation warning until 03e96ca was released in bcbio-gff v0.7.1.

This was by a test in one of my repos after the recent release of 0.7.1. Error message snippet:

+  /home/runner/micromamba/envs/augur/lib/python3.8/site-packages/Bio/Seq.py:2220: BiopythonDeprecationWarning: UnknownSeq(length) is deprecated; please use Seq(None, length) instead.

Workarounds

1. Silence deprecation warnings (similar to reverting 03e96ca).
2. Install biopython>=1.81 which uses a code path that avoids UnknownSeq:

   bcbb/gff/BCBio/GFF/GFFParser.py

   Lines 33 to 40 in 9c6d83e

   	unknown_seq_avail = False
   	try:
   	from Bio.Seq import UnknownSeq
   	unknown_seq_avail = True
   	except ImportError:
   	# Starting with biopython 1.81, has been removed
   	from Bio.Seq import _UndefinedSequenceData
   	from Bio.Seq import Seq

Potential solution

Remove references to UnknownSeq and solely use the Seq approach. It should be available with 1.79, maybe even earlier.