How to use the pyani.anib.fragment_fasta_files function in pyani

def test_blastall_graph(path_fna_all, tmp_path, fragment_length):
    """Create jobgraph for legacy BLASTN jobs."""
    fragresult = anib.fragment_fasta_files(path_fna_all, tmp_path, fragment_length)
    blastcmds = anib.make_blastcmd_builder("ANIblastall", tmp_path)
    jobgraph = anib.make_job_graph(path_fna_all, fragresult[0], blastcmds)
    # We check that the main script job is a blastn job, and that there
    # is a single dependency, which is a makeblastdb job
    for job in jobgraph:
        assert job.script.startswith("blastall -p blastn")
        assert len(job.dependencies) == 1
        assert job.dependencies[0].script.startswith("formatdb")

def test_aniblastall_concordance(
    paths_concordance_fna,
    path_concordance_jspecies,
    tolerance_anib_hi,
    fragment_length,
    tmp_path,
):
    """Check ANIblastall results are concordant with JSpecies."""
    # Get lengths of input genomes
    orglengths = pyani_files.get_sequence_lengths(paths_concordance_fna)

    # Perform ANIblastall on the input directory contents
    fragfiles, fraglengths = anib.fragment_fasta_files(
        paths_concordance_fna, tmp_path, fragment_length
    )
    jobgraph = anib.make_job_graph(
        paths_concordance_fna,
        fragfiles,
        anib.make_blastcmd_builder("ANIblastall", tmp_path),
    )
    assert 0 == run_mp.run_dependency_graph(jobgraph)  # Jobs must run correctly

    # Process BLAST output
    result_pid = anib.process_blast(
        tmp_path, orglengths, fraglengths, mode="ANIblastall"
    ).percentage_identity

    # Compare JSpecies output to results
    result_pid = (result_pid.sort_index(axis=0).sort_index(axis=1) * 100.0).values

def test_blastn_graph(path_fna_all, tmp_path, fragment_length):
    """Create jobgraph for BLASTN+ jobs."""
    fragresult = anib.fragment_fasta_files(path_fna_all, tmp_path, fragment_length)
    blastcmds = anib.make_blastcmd_builder("ANIb", tmp_path)
    jobgraph = anib.make_job_graph(path_fna_all, fragresult[0], blastcmds)
    # We check that the main script job is a blastn job, and that there
    # is a single dependency, which is a makeblastdb job
    for job in jobgraph:
        assert job.script.startswith("blastn")
        assert len(job.dependencies) == 1
        assert job.dependencies[0].script.startswith("makeblastdb")

def test_fragment_files(path_fna_all, tmp_path, dir_tgt_fragments, fragment_length):
    """Fragment files for ANIb/ANIblastall."""
    result = anib.fragment_fasta_files(path_fna_all, tmp_path, fragment_length)
    # Test that files are created
    for outfname in result[0]:
        assert outfname.is_file()

    # Test that file contents are as expected
    for fname in tmp_path.iterdir():
        with fname.open("r") as ofh:
            with (dir_tgt_fragments / fname.name).open("r") as tfh:
                assert ofh.read() == tfh.read()

    # # Test fragment lengths are in bounds
    for _, fragdict in result[-1].items():
        for _, fraglen in fragdict.items():
            assert fraglen <= fragment_length

tolerance_anib_lo,
    threshold_anib_lo_hi,
    fragment_length,
    tmp_path,
):
    """Check ANIb results are concordant with JSpecies.

    We expect ANIb results to be quite different, as the BLASTN
    algorithm changed substantially between BLAST and BLAST+ (the
    megaBLAST algorithm is now the default for BLASTN)
    """
    # Get lengths of input genomes
    orglengths = pyani_files.get_sequence_lengths(paths_concordance_fna)

    # Build and run BLAST jobs
    fragfiles, fraglengths = anib.fragment_fasta_files(
        paths_concordance_fna, tmp_path, fragment_length
    )
    jobgraph = anib.make_job_graph(
        paths_concordance_fna, fragfiles, anib.make_blastcmd_builder("ANIb", tmp_path)
    )
    assert 0 == run_mp.run_dependency_graph(jobgraph)  # Jobs must run correctly

    # Process BLAST output
    result_pid = anib.process_blast(
        tmp_path, orglengths, fraglengths, mode="ANIb"
    ).percentage_identity

    # Compare JSpecies output to results. We do this in two blocks,
    # masked according to whether the expected result is greater than
    # a threshold separating "low" from "high" identity comparisons.
    result_pid = result_pid.sort_index(axis=0).sort_index(axis=1) * 100.0