def test_compute_aa_composition_result_complex_heteropolymer():
protein_sequence = 'AWGY'
aa_composition = compute_aa_composition(protein_sequence)
assert aa_composition['A'] == 0.25
assert aa_composition['W'] == 0.25
assert aa_composition['G'] == 0.25
assert aa_composition['Y'] == 0.25
def predict_protein_location(protein_sequence:str) -> str:
aa_composition = compute_aa_composition(protein_sequence)
df_aa_composition = pd.DataFrame([aa_composition])
prediction = model.predict(df_aa_composition)
if prediction == 1:
protein_location = 'Membrane'
else:
protein_location = 'Cytoplasm'
return protein_location
def run_model(file_path: str, model_path: str) -> pd.DataFrame:
"""
Run a membrane protein prediction on a FASTA file.
Parameters
----------
file_path:str
path to proteins in FASTA format.
model_path:str
path to trained model in pickle format
Returns
-------
df_prediction:pd.DataFrame
Pandas DataFrame containing the membrane protein predictions.
"""
with open(model_path, 'rb') as handle:
model = pickle.load(handle)
handle = open(file_path)
parser = SeqIO.parse(handle, 'fasta')
df_aa_composition = pd.DataFrame()
df_predictions = pd.DataFrame(columns=['id', 'membrane'])
for record in parser:
aa_composition = compute_aa_composition(str(record.seq))
aa_composition['id'] = record.id
df_aa_composition = df_aa_composition.append(aa_composition,
ignore_index=True)
X = df_aa_composition.drop(['id'], axis=1)
ids = df_aa_composition['id']
y_pred = model.predict(X)
df_predictions['id'] = ids
df_predictions['membrane'] = y_pred
return df_predictions
def test_compute_aa_composition_result_simple_homopolymer():
protein_sequence = 'AAAAAA'
aa_composition = compute_aa_composition(protein_sequence)
assert aa_composition['A'] == 1
def test_compute_aa_composition_return_type():
protein_sequence = 'AWGY'
aa_composition = compute_aa_composition(protein_sequence)
assert isinstance(aa_composition, dict)