def test_broken_fasta(tmp_path: Path):
"""Ensure that we print a reasonable message when the user feeds in a broken fasta file.
Unfortunately, we can't detect if the user fed in a markdown file
instead of a fasta, because we could parse that markdown file
as fasta:
> Following the initial line (used for a unique description of the
> sequence) was the actual sequence itself in standard one-letter
> character string. Anything other than a valid character would
> be ignored (including spaces, tabulators, asterisks, etc...).
> It was also common to end the sequence with an "*" (asterisk)
> character (in analogy with use in PIR formatted sequences) and,
> for the same reason, to leave a blank line between the description
> and the sequence.
From https://www.wikiwand.com/en/FASTA_format
NCBI is a bit stricter with their definition by barring blank line,
but is otherwise still extremely lenient.
(https://blast.ncbi.nlm.nih.gov/Blast.cgi?CMD=Web&PAGE_TYPE=BlastDocs&DOC_TYPE=BlastHelp)
"""
input_file = "test-data/embeddings.npz"
with pytest.raises(ValueError, match="Are you sure this is a valid fasta file?"):
_process_fasta_file(
sequences_file=input_file,
prefix=str(tmp_path),
)
def test_missing_fasta(tmp_path: Path):
input_file = tmp_path.joinpath("non_existant.fasta")
with pytest.raises(FileNotFoundError, match="No such file or directory"):
_process_fasta_file(
sequences_file=input_file,
prefix=str(tmp_path),
)
def test_unparsable_fasta(caplog, pytestconfig, tmp_path: Path):
input_file = pytestconfig.rootpath.joinpath("test-data/unparsable.fasta")
with pytest.raises(
ValueError,
match=
f"Could not parse '{input_file}'. Are you sure this is a valid fasta file?",
):
_process_fasta_file(sequences_file=str(input_file),
prefix=str(tmp_path))
assert caplog.messages == []
def test_illegal_amino_acids(caplog, pytestconfig, tmp_path: Path):
"""https://github.com/sacdallago/bio_embeddings/issues/54"""
input_file = pytestconfig.rootpath.joinpath(
"test-data/illegal_amino_acids.fasta")
_process_fasta_file(sequences_file=str(input_file), prefix=str(tmp_path))
assert caplog.messages == [
f"The entry 'lowercase' in {input_file} contains lower "
"case amino acids. Lower case letters are uninterpretable by most language "
"models, and their embedding will be nonesensical. Protein LMs available "
"through bio_embeddings have been trained on upper case, single letter code "
"sequence representations only "
"(https://en.wikipedia.org/wiki/Amino_acid#Table_of_standard_amino_acid_abbreviations_and_properties)."
]
def test_illegal_amino_acids(caplog, tmp_path: Path):
""" https://github.com/sacdallago/bio_embeddings/issues/54 """
input_file = "test-data/illegal_amino_acids.fasta"
with pytest.raises(
ValueError,
match=f"The entry 'illegal' in {input_file} contains the characters 'ä', 'ö', "
"while only single letter code is allowed",
):
_process_fasta_file(
sequences_file=input_file,
prefix=str(tmp_path),
)
assert caplog.messages == [
f"The entry 'lowercase' in {input_file} contains lower "
"case amino acids. Lower case letters are uninterpretable by most language "
"models, and their embedding will be nonesensical. Protein LMs available "
"through bio_embeddings have been trained on upper case, single letter code "
"sequence representations only "
"(https://en.wikipedia.org/wiki/Amino_acid#Table_of_standard_amino_acid_abbreviations_and_properties)."
]
def test_simple_remapping(tmp_path: Path):
""" https://github.com/sacdallago/bio_embeddings/issues/50 """
global_parameters = {
"sequences_file": "test-data/seqwence-protein.fasta",
"prefix": str(tmp_path),
"simple_remapping": True,
}
global_parameters = _process_fasta_file(**global_parameters)
embed_and_write_batched(
FakeEmbedder(),
FileSystemFileManager(),
global_parameters,
)
def test_simple_remapping():
""" https://github.com/sacdallago/bio_embeddings/issues/50 """
with TemporaryDirectory() as prefix:
global_parameters = {
"sequences_file": "test-data/seqwence-protein.fasta",
"prefix": prefix,
"simple_remapping": True,
}
global_parameters = _process_fasta_file(**global_parameters)
embed_and_write_batched(
FakeEmbedder(),
FileSystemFileManager(),
{
**global_parameters,
"max_amino_acids": 10000,
"discard_per_amino_acid_embeddings": False,
},
)
