def output(self):
output_paths = []
for model in self.models:
output_paths.append("{}_pca.csv".format(model))
outputs = [paths.output(p, 'pca') for p in output_paths]
return [luigi.LocalTarget(output) for output in outputs]
def run(self):
metadata = pd.read_csv(self.input()[0].fn, index_col=0)
metadata['sample_id'] = metadata['sample_id'].astype(str)
metadata = metadata.set_index('sample_id')
metadata = metadata[['env_material']]
metrics = pd.read_csv(self.input()[1][0].fn)
for i, model in enumerate(self.models):
df = metrics[metrics['embedding'] == model]
best = df.loc[df['test_f1_score'].idxmax()]
best_training_data = best['training_data_name']
training_data_path = paths.output(best_training_data,
'training_data')
training_data = pd.read_pickle(training_data_path)
print('Training data shape', training_data.shape)
idx = training_data.index
X = training_data.drop(self.target, axis=1)
pca = PCA(n_components=3)
transformed = pd.DataFrame(pca.fit_transform(X), index=idx)
transformed = transformed.merge(metadata,
left_index=True,
right_index=True)
print("{} Explained Variance".format(model),
pca.explained_variance_ratio_)
output_path = self.output()[i].path
transformed.to_csv(output_path)
def output(self):
output_paths = [
'speciesid_to_tax.csv', 'taxonomy_97_transitive_closure.csv'
]
outputs = [paths.output(p) for p in output_paths]
return [luigi.LocalTarget(output) for output in outputs]
def output(self):
output_paths = [
("{}_lr_model.pkl".format(self.name()), 'model'),
("{}_lr_model_metrics.csv".format(self.name()), 'metrics'),
(self.training_data_name(), 'training_data'),
]
outputs = [paths.output(p[0], p[1]) for p in output_paths]
return [luigi.LocalTarget(output) for output in outputs]
def output(self):
filename = "biom.pkl"
local_file_path = paths.output(filename)
return luigi.LocalTarget(local_file_path)
def output(self):
filename = "alpha_diversity.pkl"
local_file_path = paths.output(filename)
return luigi.LocalTarget(local_file_path)
def output(self):
local_file_path = paths.output(self.filename)
return luigi.LocalTarget(local_file_path)
def output(self):
output_paths = ['labeled_metadata.csv', 'label_statistics.csv']
outputs = [paths.output(p) for p in output_paths]
return [luigi.LocalTarget(output) for output in outputs]
def output(self):
output_paths = ['body_site.csv']
outputs = [paths.output(p) for p in output_paths]
return [luigi.LocalTarget(output) for output in outputs]
def output(self):
filename = "{}_training_data.pkl".format(self.target)
local_file_path = paths.output(filename)
return luigi.LocalTarget(local_file_path)
def output(self):
output_paths = ["{}_combined_metrics.csv".format(self.target)]
outputs = [paths.output(p, 'metrics') for p in output_paths]
return [luigi.LocalTarget(output) for output in outputs]
def output(self):
filename = "sentences_{}.cor".format(self.use_value)
local_file_path = paths.output(filename)
return luigi.LocalTarget(local_file_path)
def output(self):
filename = "biom_dim_w2v_{}_{}_{}_{}.pkl".format(
self.use_value, self.min_count, self.size, self.epochs)
local_file_path = paths.output(filename)
return luigi.LocalTarget(local_file_path)
def output(self):
output_paths = ['sample_id_to_tax.csv', 'hyperbolic_df.pkl']
outputs = [paths.output(p) for p in output_paths]
return [luigi.LocalTarget(output) for output in outputs]
