Perform protein structure-based sequence alignment in Python

Question

I am looking for a Python package that performs pairwise structural alignment of protein structures (i.e., PDB files) and returns a sequence alignment. PyMOL is able to do this through the GUI, for example:

For two protein PDBs, one can be aligned to the other with the "align" menu or align command:

With it, the sequences appear aligned in the sequence viewer according to the structural alignment, and an "alignment" object is created, which can be saved with the command:

save A_aligned_with_B.aln, aln_structure_1_to_structure_0

which produces the A_aligned_with_B.aln alignment file in Clustal format (nevermind the ?s, which correspond to special HIS resnames):

CLUSTAL

structure_0  YIYQDSISLPWKVLLVMLLALITLATTLSNAFVIATVYRTRKL?TPANYLIASLAVTDLLVSI
structure_1  EEQGNKL?WAALLILMVIIPTIG-----GNTLVILAVSLEKKLQYATNYFLMSLAVADLLVGL
                                         .*..**..*....**....**...****.****..

structure_0  LVMPISTMYTVT-GRWTLGQVVCDFWLSSDITCCTASIL?LCVIALDRYWAITDAVEYSAKRT
structure_1  FVMPIALLTIMFEAMWPLPLVLCPAWLFLDVLFSTASIM?LCAISVDRYIAIKKPIQANQYNS
             .****......  ..*.*..*.*..**..*....****.***.*..***.**........ ..

structure_0  PKRAAVMIALVWVFSISISLPP-FFWRQAKAEEEVSE-CVVNTD?IL-YTVYSTVGAFYFPTL
structure_1  RATAFIKITVVWLISIGIAIPVPIKGIETDVDNPNNITCVLTKERFGDFMLFGSLAAFFTPLA
             ...*...*..**..**.*..*  .              **....    ........**..*..

structure_0  LLIALYGRIYVEARSRILKQTPNRTGKLLEKKKLMAAR--ERKATKTLGIILGAFIVCWLPFF
structure_1  IMIVTYFLTI?ALQKKAYLVKNKPPQRLTIGKKSVQTISNEQRASKVLGIVFFLFLLMWCPFF
             ..*..*....                              * .*.*.***....*...*.***

structure_0  IISLVMPICKDACW--F?LAIFDFFTWLGYLNSLINPIIYTMSNEDFKQAF?KLIRFK--   
structure_1  ITNITLVLCDS-CNQTTLQMLLEIFVWIGYVSSGVNPLVYTLFNKTFRDAFGRYITCNYR   
             *.......*   *   .  .....*.*.**..*..**..**..*..*..**         

I need a Python package that is easily installable/distributable that accepts two PDBs or protein structures and returns the structural alignment of the sequences somehow. All within a Python script, and not creating intermediate files, e.g. for the alignment.

There are plenty of Python tools that allow structural superimposition and RMSD calculation (e.g., Biopython), but I haven't found any that are able to return a sequence alignment from the structural alignment.

Edit:

Possible answer, thanks to the answer and comments from Matteo Ferla.

import pymol2
with pymol2.PyMOL() as pymol:
    pymol.cmd.load('foo.pdb', 'foo')
    pymol.cmd.load('bar.pdb', 'bar')
    pymol.cmd.align('foo', 'bar', object='aln')
    aln = pymol2.cmd2.pymol.exporting.get_alnstr('aln', _self=pymol.cmd) # I don't know if this line could be cleaner
    # aln is itself a string

Answer 1

PyMOL in itself can be used as python package. The GUI app has its own Python, but in a conda environment you can install it as a package. Any PyMOL app command can be used as a cmd command. In PyMOL Wiki you see both. They follow a similar pattern and actually have more functionality.

import pymol2
with pymol2.PYMOL() as pymol:
    pymol.cmd.load('foo.pdb', 'foo')
    pymol.cmd.load('bar.pdb', 'bar')
    pymol.cmd.align('foo', 'bar')
    pymol.cmd.save('foobar.aln')

For more discussion of how the alignment output object is used see Assessing PyMol sequence alignment object