Changelog

Additions

• Support for Python 3.13 (#661).
• Support for structural alphabets that encodes geometric information from structure.AtomArray objects in sequence.Sequence objects.
  • structure.alphabet.to_3di converts a structure into a 3Di sequence from Foldseek. (#665)
  • structure.alphabet.to_protein_blocks() converts a structure into a Protein Blocks sequence. (#676)
  • Added color schemes and default sequence.align.SubstitutionMatrix for these structural alphabets. (#682)
• Support for positional substitution matrices. (#655)
  • sequence.PositionalSequence acts as a placeholder sequence for a sequence profile in alignment functions from sequence.align.
  • sequence.align.SubstitutionMatrix.as_positional() expands a substitution matrix into a positional substitution matrix.
• New functionalities for structure.io.pdbx:
  • Custom annotations can be written with set_structure() with extra_fields=True. (#669)
  • Missing columns in atom_site category are now handled by using sensible default annotations. (#670)
  • Added compress() function that automatically finds optimal encodings for BinaryCIFFile, decreasing the file size by a factor of approx 8x. (#674)
  • get_assembly() now adds a sym_id annotation to better distinguish copies of the asymmetric unit. (#700)
• Added PDBFile.get_space_group() and PDBFile.set_space_group() to read and write the space group information. (#707)
• Added structure.concatenate(), which supports concatenating more than AtomArrayStack objects. (#712)
• Performance improvements:
  • Doubled speed to parsing CIF files with structure.io.pdbx.CIFFile (#722).
  • Tripled speed of pickling sequence.align.KmerTable objects. (#664)
    • Speed of transferring sequence.align.KmerTable between processes improves likewise.

Changes

• structure.io.pdbx.set_structure() does not write 'canonical' bonds to struct_conn category anymore. (#678)
• The internal Chemical Component Dictionary is not version controlled anymore. (#687, #716)
  • For installations from local repository clones, it must be built with python -m biotite.setup_ccd.

Fixes

• structure.io.pdbx can now handle atom_site categories with quoted values containing whitespaces. (#673)
• nan values in structure.AtomArray now count as equal when comparing AtomArray objects. (#714)
• Fixed wrong band calculation in sequence.align.align_banded() that occurred when the given band was outside the sequence bounds. (#723)
  • This lead to premature traceback termination in rare cases.