def test_ncbi_stockholm(
namespace_base,
path_ncbi_stockholm,
path_ncbi_stockholm_targets,
tmp_path,
):
"""ncfp collects correct coding sequences for NCBI/Stockholm input.
This test was added as a check for the fix in issue 31.
"""
# Modify default arguments
infile = path_ncbi_stockholm
outdir = tmp_path / "ncbi_stockholm"
args = modify_namespace(namespace_base,
infname=infile,
outdirname=outdir,
stockholm=True)
# Run ersatz command-line
ncfp.run_main(args)
# Compare output (should be no skipped files)
check_files(
outdir,
path_ncbi_stockholm_targets,
("ncfp_aa.fasta", "ncfp_nt.fasta"),
)
def test_small_stockholm_use_protein_id(
namespace_base,
path_uniprot_stockholm_small,
path_uniprot_stockholm_small_use_proteinid_targets,
tmp_path,
):
"""ncfp collects correct coding sequences for small UniProt/Stockholm input."""
# Modify default arguments
infile = path_uniprot_stockholm_small
outdir = tmp_path / "small_stockholm_use_proteinid"
args = modify_namespace(
namespace_base,
infname=infile,
outdirname=outdir,
stockholm=True,
use_protein_ids=True,
)
# Run ersatz command-line
ncfp.run_main(args)
# Compare output (should be no skipped files)
check_files(
outdir,
path_uniprot_stockholm_small_use_proteinid_targets,
("ncfp_aa.fasta", "ncfp_nt.fasta"),
)
def test_bad_infile(args_bad_infile):
"""ncfp stops if CLI input file does not exist.
Validates that ncfp throws correct error, does not check output
"""
with pytest.raises(NCFPException):
ncfp.run_main(args_bad_infile)
def test_download_and_log(args_validate_and_log,
mock_uniprot_download_and_log):
"""ncfp downloads coding sequences and logs output from CLI.
Validates that ncfp runs without error, does not check output. Calls to UniProt
are mocked for use with cloud CI.
"""
ncfp.run_main(args_validate_and_log)
def test_create_and_keep_cache(args_create_reuse_cache,
mock_entrez_create_and_keep_cache):
"""ncfp creates named cache from CLI and keeps it when rerunning.
This calls ncfp twice and expects both calls run without error. Output
is not checked.
"""
# Create local cache
ncfp.run_main(args_create_reuse_cache)
# Reuse created cache
ncfp.run_main(args_create_reuse_cache + ["--keepcache"])
def test_basic_ncbi(namespace_base, path_ncbi, path_ncbi_targets, tmp_path):
"""ncfp collects correct coding sequences for basic NCBI input."""
# Modify default arguments
infile = path_ncbi
outdir = tmp_path / "basic_ncbi"
args = modify_namespace(namespace_base, infname=infile, outdirname=outdir)
# Run ersatz command-line
ncfp.run_main(args)
# Compare output
check_files(outdir, path_ncbi_targets, ("ncfp_aa.fasta", "ncfp_nt.fasta"))
def test_alternative_start(namespace_base, path_altstart,
path_altstart_targets, tmp_path):
"""ncfp collects correct coding sequences for NCBI input with alternative start codon."""
infile = path_altstart
outdir = tmp_path / "alternative_start"
args = modify_namespace(namespace_base,
infname=infile,
outdirname=outdir,
alternative_start_codon=True)
# Run ersatz command-line
ncfp.run_main(args)
# Compare output
check_files(outdir, path_altstart_targets,
("ncfp_aa.fasta", "ncfp_nt.fasta"))
def test_basic_uniprot(namespace_base, path_uniprot, path_uniprot_targets,
tmp_path, mock_basic_uniprot):
"""ncfp collects correct coding sequences for basic UniProt input."""
# Modify default arguments
infile = path_uniprot
outdir = tmp_path / "basic_uniprot"
args = modify_namespace(namespace_base, infname=infile, outdirname=outdir)
# Run ersatz command-line
ncfp.run_main(args)
# Compare output
check_files(
outdir,
path_uniprot_targets,
("ncfp_aa.fasta", "ncfp_nt.fasta", "skipped.fasta"),
)
def test_basic_stockholm(namespace_base, path_stockholm,
path_stockholm_targets, tmp_path):
"""ncfp collects correct coding sequences for basic UniProt/Stockholm input."""
# Modify default arguments
infile = path_stockholm
outdir = tmp_path / "basic_stockholm"
args = modify_namespace(namespace_base,
infname=infile,
outdirname=outdir,
stockholm=True)
# Run ersatz command-line
ncfp.run_main(args)
# Compare output (should be no skipped files)
check_files(
outdir,
path_stockholm_targets,
("ncfp_aa.fasta", "ncfp_nt.fasta"),
)