def pipeline(data_file,
cords_file,
results_dir,
n_components=50,
n_components_plot=5):
"""
Pipeline for generating PCA plots from data file containing features.
:param data_file: Data file containing features matrix. Must have one
column with labels named "label" and features in columns "aaaaaa" to
"tttttt" in header. Include directory and extension.
:param cords_file: File to store transformed coordinates to, including
directory and extension.
:param results_dir: Directory to store scatter plots to.
:param n_components: Number of components to reduce to (default = 50).
:param n_components_plot: Number of principal components to generate
combinations to plot (default = 5).
:return: None.
"""
# Load in data file
print("Loading data file...")
data_frame = pd.read_table(data_file)
# Get features used in PCA
features_df = data_frame.loc[:, "aaaaaa":"tttttt"]
# Get labels
labels_df = data_frame.loc[:, "label"]
labels_list = sorted(list(set(labels_df)))
# Create the PCA
print("Creating PCA...")
ipca = IncrementalPCA(n_components=n_components)
features_transformed = ipca.fit_transform(features_df)
# Label the transformed coordinates
transformed_df = pca.label_coordinates(
transformed_coordinates=features_transformed, labels=labels_df)
# Save the transformed coordinates
print("Saving transformed coordinates...")
transformed_df.to_csv(cords_file, sep='\t', index=False)
# Save the explained variances
print("Saving explained variances...")
pca.save_variances(pca=ipca, file_name=results_dir + "variances.txt")
# Create combinations of principal components to plot
components = range(1, n_components_plot + 1)
combinations_list = generate_combinations(components)
# Plot all the combinations of components. Choose colors list with colors
# in matplotlib.
plot_combinations(transformed_df,
results_dir=results_dir,
labels_list=labels_list,
combinations_list=combinations_list,
colors_list=[
"saddlebrown", "gold", "darkgreen", "cyan",
"darkblue", "magenta"
],
ipca=ipca)
def test_generate_combinations_4():
assert sp.generate_combinations("abcd", n = 3) == [('a', 'b', 'c'), ('a', 'b', 'd'),
('a', 'c', 'd'), ('b', 'c', 'd')]
def test_generate_combinations_3():
assert sp.generate_combinations("ab", n = 3) == []
def test_generate_combinations_2():
assert sp.generate_combinations([1, 2, 3]) == [(1, 2), (1, 3), (2, 3)]
def test_generate_combinations_1():
assert sp.generate_combinations([]) == []