def test_encode(self):
path = EnvironmentSettings.tmp_test_path + "abundance_encoder/"
PathBuilder.build(path)
repertoires, metadata = RepertoireBuilder.build([["GGG", "III", "LLL", "MMM"],
["DDD", "EEE", "FFF", "III", "LLL", "MMM"],
["CCC", "FFF", "MMM"],
["AAA", "CCC", "EEE", "FFF", "LLL", "MMM"]],
labels={"l1": [True, True, False, False]}, path=path)
dataset = RepertoireDataset(repertoires=repertoires, metadata_file=metadata, identifier="1")
encoder = SequenceAbundanceEncoder.build_object(dataset, **{
"comparison_attributes": ["sequence_aas"],
"p_value_threshold": 0.4, "sequence_batch_size": 4, "repertoire_batch_size": 8
})
label_config = LabelConfiguration([Label("l1", [True, False], positive_class=True)])
encoded_dataset = encoder.encode(dataset, EncoderParams(result_path=path, label_config=label_config))
self.assertTrue(np.array_equal(np.array([[1, 4], [1, 6], [0, 3], [0, 6]]), encoded_dataset.encoded_data.examples))
encoder.p_value_threshold = 0.05
encoded_dataset = encoder.encode(dataset, EncoderParams(result_path=path, label_config=label_config))
self.assertTrue(np.array_equal(np.array([[0, 4], [0, 6], [0, 3], [0, 6]]), encoded_dataset.encoded_data.examples))
shutil.rmtree(path)
def import_hp_setting(config_dir: str) -> Tuple[HPSetting, Label]:
config = MLMethodConfiguration()
config.load(f'{config_dir}ml_config.yaml')
ml_method = ReflectionHandler.get_class_by_name(
config.ml_method, 'ml_methods/')()
ml_method.load(config_dir)
encoder = MLImport.import_encoder(config, config_dir)
preprocessing_sequence = MLImport.import_preprocessing_sequence(
config, config_dir)
labels = list(config.labels_with_values.keys())
assert len(
labels
) == 1, "MLImport: Multiple labels set in a single ml_config file."
label = Label(labels[0], config.labels_with_values[labels[0]])
return HPSetting(
encoder=encoder,
encoder_params=config.encoding_parameters,
encoder_name=config.encoding_name,
ml_method=ml_method,
ml_method_name=config.ml_method_name,
ml_params={},
preproc_sequence=preprocessing_sequence,
preproc_sequence_name=config.preprocessing_sequence_name), label
def test_sequence_flattened(self):
path = EnvironmentSettings.root_path + "test/tmp/onehot_seq_flat/"
PathBuilder.build(path)
dataset = self.construct_test_flatten_dataset(path)
encoder = OneHotEncoder.build_object(dataset, **{"use_positional_info": False, "distance_to_seq_middle": None, "flatten": True})
encoded_data = encoder.encode(dataset, EncoderParams(
result_path=path,
label_config=LabelConfiguration([Label(name="l1", values=[1, 0], positive_class="1")]),
pool_size=1,
learn_model=True,
model={},
filename="dataset.pkl"
))
self.assertTrue(isinstance(encoded_data, SequenceDataset))
onehot_a = [1.0] + [0.0] * 19
onehot_t = [0.0] * 16 + [1.0] + [0] * 3
self.assertListEqual(list(encoded_data.encoded_data.examples[0]), onehot_a+onehot_a+onehot_a+onehot_t+onehot_t+onehot_t)
self.assertListEqual(list(encoded_data.encoded_data.examples[1]), onehot_a+onehot_t+onehot_a+onehot_t+onehot_a+onehot_t)
self.assertListEqual(list(encoded_data.encoded_data.feature_names), [f"{pos}_{char}" for pos in range(6) for char in EnvironmentSettings.get_sequence_alphabet()])
shutil.rmtree(path)
def _create_state_object(self, path):
repertoires, metadata = RepertoireBuilder.build(sequences=[["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"]],
path=path,
labels={
"l1": [1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2,
1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2],
"l2": [0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 1,
0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1]})
dataset = RepertoireDataset(repertoires=repertoires, metadata_file=metadata,
params={"l1": [1, 2], "l2": [0, 1]})
enc_params = {"k": 3, "model_type": ModelType.SEQUENCE.name, "vector_size": 4}
hp_settings = [HPSetting(Word2VecEncoder.build_object(dataset, **enc_params), enc_params,
LogisticRegression(),
{"model_selection_cv": False, "model_selection_n_folds": -1},
[])]
label_config = LabelConfiguration([Label("l1", [1, 2]), Label("l2", [0, 1])])
process = TrainMLModelInstruction(dataset, GridSearch(hp_settings), hp_settings,
SplitConfig(SplitType.RANDOM, 1, 0.5),
SplitConfig(SplitType.RANDOM, 1, 0.5),
{Metric.BALANCED_ACCURACY}, Metric.BALANCED_ACCURACY, label_config, path)
state = process.run(result_path=path)
return state
def test_run(self):
path = EnvironmentSettings.tmp_test_path + "mlapplicationtest/"
PathBuilder.build(path)
dataset = RandomDatasetGenerator.generate_repertoire_dataset(
50, {5: 1}, {5: 1}, {"l1": {
1: 0.5,
2: 0.5
}}, path + 'dataset/')
ml_method = LogisticRegression()
encoder = KmerFreqRepertoireEncoder(
NormalizationType.RELATIVE_FREQUENCY,
ReadsType.UNIQUE,
SequenceEncodingType.CONTINUOUS_KMER,
3,
scale_to_zero_mean=True,
scale_to_unit_variance=True)
label_config = LabelConfiguration([Label("l1", [1, 2])])
enc_dataset = encoder.encode(
dataset,
EncoderParams(result_path=path,
label_config=label_config,
filename="tmp_enc_dataset.pickle",
pool_size=4))
ml_method.fit(enc_dataset.encoded_data, 'l1')
hp_setting = HPSetting(
encoder, {
"normalization_type": "relative_frequency",
"reads": "unique",
"sequence_encoding": "continuous_kmer",
"k": 3,
"scale_to_zero_mean": True,
"scale_to_unit_variance": True
}, ml_method, {}, [], 'enc1', 'ml1')
PathBuilder.build(path + 'result/instr1/')
shutil.copy(path + 'dict_vectorizer.pickle',
path + 'result/instr1/dict_vectorizer.pickle')
shutil.copy(path + 'scaler.pickle',
path + 'result/instr1/scaler.pickle')
ml_app = MLApplicationInstruction(dataset, label_config, hp_setting, 4,
"instr1", False)
ml_app.run(path + 'result/')
predictions_path = path + "result/instr1/predictions.csv"
self.assertTrue(os.path.isfile(predictions_path))
df = pd.read_csv(predictions_path)
self.assertEqual(50, df.shape[0])
shutil.rmtree(path)
def test_encode(self):
path = EnvironmentSettings.tmp_test_path + "distance_encoder/"
PathBuilder.build(path)
dataset = self.create_dataset(path)
enc = DistanceEncoder.build_object(
dataset, **{
"distance_metric": DistanceMetricType.JACCARD.name,
"attributes_to_match": ["sequence_aas"],
"sequence_batch_size": 20
})
enc.set_context({"dataset": dataset})
encoded = enc.encode(
dataset,
EncoderParams(result_path=path,
label_config=LabelConfiguration(
[Label("l1", [0, 1]),
Label("l2", [2, 3])]),
pool_size=4,
filename="dataset.pkl"))
self.assertEqual(8, encoded.encoded_data.examples.shape[0])
self.assertEqual(8, encoded.encoded_data.examples.shape[1])
self.assertEqual(1, encoded.encoded_data.examples.iloc[0, 0])
self.assertEqual(1, encoded.encoded_data.examples.iloc[1, 1])
self.assertEqual(1, encoded.encoded_data.examples.iloc[0, 4])
self.assertTrue(
np.array_equal([1, 0, 1, 0, 1, 0, 1, 0],
encoded.encoded_data.labels["l1"]))
self.assertTrue(
np.array_equal([2, 3, 2, 3, 2, 3, 3, 3],
encoded.encoded_data.labels["l2"]))
shutil.rmtree(path)
def test_encode(self):
path = EnvironmentSettings.tmp_test_path + "deeprc_encoder/"
PathBuilder.build(path)
PathBuilder.build(path + "encoded_data/")
main_dataset, sub_dataset = self.create_datasets(path)
enc = DeepRCEncoder.build_object(sub_dataset, **{})
enc.set_context({"dataset": main_dataset})
encoded = enc.encode(
sub_dataset,
EncoderParams(result_path=path + "encoded_data/",
label_config=LabelConfiguration(
[Label("l1", [0, 1]),
Label("l2", [2, 3])]),
pool_size=4))
self.assertListEqual(encoded.encoded_data.example_ids,
sub_dataset.get_repertoire_ids())
self.assertTrue(
os.path.isfile(encoded.encoded_data.info["metadata_filepath"]))
metadata_content = pd.read_csv(
encoded.encoded_data.info["metadata_filepath"], sep="\t")
self.assertListEqual(list(metadata_content["ID"]),
sub_dataset.get_repertoire_ids())
for repertoire in main_dataset.repertoires:
rep_path = f"{path}/encoded_data/encoding/{repertoire.identifier}.tsv"
self.assertTrue(os.path.isfile(rep_path))
repertoire_tsv = pd.read_csv(rep_path, sep="\t")
self.assertListEqual(list(repertoire_tsv["amino_acid"]),
list(repertoire.get_sequence_aas()))
shutil.rmtree(path)
def test_generate(self):
path = PathBuilder.build(EnvironmentSettings.tmp_test_path +
"tcrdist_motif_discovery/")
dataset_path = self._create_dataset(path)
dataset = SingleLineReceptorImport.import_dataset(
{
"path":
dataset_path,
"result_path":
path + "dataset/",
"separator":
",",
"columns_to_load": [
"subject", "epitope", "count", "v_a_gene", "j_a_gene",
"cdr3_a_aa", "v_b_gene", "j_b_gene", "cdr3_b_aa",
"clone_id", "cdr3_a_nucseq", "cdr3_b_nucseq"
],
"column_mapping": {
"cdr3_a_aa": "alpha_amino_acid_sequence",
"cdr3_b_aa": "beta_amino_acid_sequence",
"cdr3_a_nucseq": "alpha_nucleotide_sequence",
"cdr3_b_nucseq": "beta_nucleotide_sequence",
"v_a_gene": "alpha_v_gene",
"v_b_gene": "beta_v_gene",
"j_a_gene": "alpha_j_gene",
"j_b_gene": "beta_j_gene",
"clone_id": "identifier"
},
"receptor_chains":
"TRA_TRB",
"region_type":
"IMGT_CDR3",
"sequence_file_size":
50000,
"organism":
"mouse"
}, 'd1')
dataset = TCRdistEncoder(8).encode(
dataset,
EncoderParams(f"{path}result/",
LabelConfiguration([Label("epitope")])))
report = TCRdistMotifDiscovery(dataset, path + "report/",
"report name", 8)
report.generate_report()
shutil.rmtree(path)
def encode_dataset_by_kmer_freq(path_to_dataset_directory: str, result_path: str, metadata_path: str = None):
"""
encodes the repertoire dataset using KmerFrequencyEncoder
:param path_to_dataset_directory: path to directory containing all repertoire files with .tsv extension in MiXCR format
:param result_path: where to store the results
:param metadata_path: csv file with columns "filename", "subject_id", "disease" which is filled by default if value of argument is None,
otherwise any metadata csv file passed to the function, must include filename and subject_id columns,
and an arbitrary disease column
:return: encoded dataset with encoded data in encoded_dataset.encoded_data.examples
"""
if metadata_path is None:
metadata_path = generate_random_metadata(path_to_dataset_directory, result_path)
loader = MiXCRImport()
dataset = loader.import_dataset({
"is_repertoire": True,
"path": path_to_dataset_directory,
"metadata_file": metadata_path,
"region_type": "IMGT_CDR3", # import_dataset in only cdr3
"number_of_processes": 4, # number of parallel processes for loading the data
"result_path": result_path,
"separator": "\t",
"columns_to_load": ["cloneCount", "allVHitsWithScore", "allJHitsWithScore", "aaSeqCDR3", "nSeqCDR3"],
"column_mapping": {
"cloneCount": "counts",
"allVHitsWithScore": "v_genes",
"allJHitsWithScore": "j_genes"
},
}, "mixcr_dataset")
label_name = list(dataset.params.keys())[0] # label that can be used for ML prediction - by default: "disease" with values True/False
encoded_dataset = DataEncoder.run(DataEncoderParams(dataset, KmerFrequencyEncoder.build_object(dataset, **{
"normalization_type": "relative_frequency", # encode repertoire by the relative frequency of k-mers in repertoire
"reads": "unique", # count each sequence only once, do not use clonal count
"k": 2, # k-mer length
"sequence_encoding": "continuous_kmer" # split each sequence in repertoire to overlapping k-mers
}), EncoderParams(result_path=result_path,
label_config=LabelConfiguration([Label(label_name, dataset.params[label_name])])), False))
dataset_exporter = DesignMatrixExporter(dataset=encoded_dataset,
result_path=f"{result_path if result_path[:-1] == '/' else result_path+'/'}csv_exported/")
dataset_exporter.generate_report()
return encoded_dataset
def test_encode(self):
file_content = """complex.id Gene CDR3 V J Species MHC A MHC B MHC class Epitope Epitope gene Epitope species Reference Method Meta CDR3fix Score
3050 TRB CASSPPRVYSNGAGLAGVGWRNEQFF TRBV5-4*01 TRBJ2-1*01 HomoSapiens HLA-A*11:01 B2M MHCI AVFDRKSDAK EBNA4 EBV https://www.10xgenomics.com/resources/application-notes/a-new-way-of-exploring-immunity-linking-highly-multiplexed-antigen-recognition-to-immune-repertoire-and-phenotype/# {"frequency": "1/11684", "identification": "dextramer-sort", "sequencing": "rna-seq", "singlecell": "yes", "verification": ""} {"cell.subset": "", "clone.id": "", "donor.MHC": "", "donor.MHC.method": "", "epitope.id": "", "replica.id": "", "samples.found": 1, "structure.id": "", "studies.found": 1, "study.id": "", "subject.cohort": "", "subject.id": "1", "tissue": ""} {"cdr3": "CASSPPRVYSNGAGLAGVGWRNEQFF", "cdr3_old": "CASSPPRVYSNGAGLAGVGWRNEQFF", "fixNeeded": false, "good": true, "jCanonical": true, "jFixType": "NoFixNeeded", "jId": "TRBJ2-1*01", "jStart": 21, "vCanonical": true, "vEnd": 4, "vFixType": "NoFixNeeded", "vId": "TRBV5-4*01"} 0
15760 TRB CASSWTWDAATLWGQGALGGANVLTF TRBV5-5*01 TRBJ2-6*01 HomoSapiens HLA-A*03:01 B2M MHCI KLGGALQAK IE1 CMV https://www.10xgenomics.com/resources/application-notes/a-new-way-of-exploring-immunity-linking-highly-multiplexed-antigen-recognition-to-immune-repertoire-and-phenotype/# {"frequency": "1/25584", "identification": "dextramer-sort", "sequencing": "rna-seq", "singlecell": "yes", "verification": ""} {"cell.subset": "", "clone.id": "", "donor.MHC": "", "donor.MHC.method": "", "epitope.id": "", "replica.id": "", "samples.found": 1, "structure.id": "", "studies.found": 1, "study.id": "", "subject.cohort": "", "subject.id": "3", "tissue": ""} {"cdr3": "CASSWTWDAATLWGQGALGGANVLTF", "cdr3_old": "CASSWTWDAATLWGQGALGGANVLTF", "fixNeeded": false, "good": true, "jCanonical": true, "jFixType": "NoFixNeeded", "jId": "TRBJ2-6*01", "jStart": 19, "vCanonical": true, "vEnd": 4, "vFixType": "NoFixNeeded", "vId": "TRBV5-5*01"} 0
3050 TRA CAAIYESRGSTLGRLYF TRAV13-1*01 TRAJ18*01 HomoSapiens HLA-A*11:01 B2M MHCI AVFDRKSDAK EBNA4 EBV https://www.10xgenomics.com/resources/application-notes/a-new-way-of-exploring-immunity-linking-highly-multiplexed-antigen-recognition-to-immune-repertoire-and-phenotype/# {"frequency": "1/11684", "identification": "dextramer-sort", "sequencing": "rna-seq", "singlecell": "yes", "verification": ""} {"cell.subset": "", "clone.id": "", "donor.MHC": "", "donor.MHC.method": "", "epitope.id": "", "replica.id": "", "samples.found": 1, "structure.id": "", "studies.found": 1, "study.id": "", "subject.cohort": "", "subject.id": "1", "tissue": ""} {"cdr3": "CAAIYESRGSTLGRLYF", "cdr3_old": "CAAIYESRGSTLGRLYF", "fixNeeded": false, "good": true, "jCanonical": true, "jFixType": "NoFixNeeded", "jId": "TRAJ18*01", "jStart": 7, "oldVEnd": -1, "oldVFixType": "FailedBadSegment", "oldVId": null, "vCanonical": true, "vEnd": 3, "vFixType": "ChangeSegment", "vId": "TRAV13-1*01"} 0
15760 TRA CALRLNNQGGKLIF TRAV9-2*01 TRAJ23*01 HomoSapiens HLA-A*03:01 B2M MHCI KLGGALQAK IE1 CMV https://www.10xgenomics.com/resources/application-notes/a-new-way-of-exploring-immunity-linking-highly-multiplexed-antigen-recognition-to-immune-repertoire-and-phenotype/# {"frequency": "1/25584", "identification": "dextramer-sort", "sequencing": "rna-seq", "singlecell": "yes", "verification": ""} {"cell.subset": "", "clone.id": "", "donor.MHC": "", "donor.MHC.method": "", "epitope.id": "", "replica.id": "", "samples.found": 1, "structure.id": "", "studies.found": 1, "study.id": "", "subject.cohort": "", "subject.id": "3", "tissue": ""} {"cdr3": "CALRLNNQGGKLIF", "cdr3_old": "CALRLNNQGGKLIF", "fixNeeded": false, "good": true, "jCanonical": true, "jFixType": "NoFixNeeded", "jId": "TRAJ23*01", "jStart": 6, "vCanonical": true, "vEnd": 3, "vFixType": "NoFixNeeded", "vId": "TRAV9-2*01"} 0
3051 TRB CASSPPRVYSNGAGLAGVGWRNEQFF TRBV5-4*01 TRBJ2-1*01 HomoSapiens HLA-A*11:01 B2M MHCI AVFDRKSDAK EBNA4 EBV https://www.10xgenomics.com/resources/application-notes/a-new-way-of-exploring-immunity-linking-highly-multiplexed-antigen-recognition-to-immune-repertoire-and-phenotype/# {"frequency": "1/11684", "identification": "dextramer-sort", "sequencing": "rna-seq", "singlecell": "yes", "verification": ""} {"cell.subset": "", "clone.id": "", "donor.MHC": "", "donor.MHC.method": "", "epitope.id": "", "replica.id": "", "samples.found": 1, "structure.id": "", "studies.found": 1, "study.id": "", "subject.cohort": "", "subject.id": "1", "tissue": ""} {"cdr3": "CASSPPRVYSNGAGLAGVGWRNEQFF", "cdr3_old": "CASSPPRVYSNGAGLAGVGWRNEQFF", "fixNeeded": false, "good": true, "jCanonical": true, "jFixType": "NoFixNeeded", "jId": "TRBJ2-1*01", "jStart": 21, "vCanonical": true, "vEnd": 4, "vFixType": "NoFixNeeded", "vId": "TRBV5-4*01"} 0
15761 TRB CASSWTWDAATLWGQGALGGANVLTF TRBV5-5*01 TRBJ2-6*01 HomoSapiens HLA-A*03:01 B2M MHCI KLGGALQAK IE1 CMV https://www.10xgenomics.com/resources/application-notes/a-new-way-of-exploring-immunity-linking-highly-multiplexed-antigen-recognition-to-immune-repertoire-and-phenotype/# {"frequency": "1/25584", "identification": "dextramer-sort", "sequencing": "rna-seq", "singlecell": "yes", "verification": ""} {"cell.subset": "", "clone.id": "", "donor.MHC": "", "donor.MHC.method": "", "epitope.id": "", "replica.id": "", "samples.found": 1, "structure.id": "", "studies.found": 1, "study.id": "", "subject.cohort": "", "subject.id": "3", "tissue": ""} {"cdr3": "CASSWTWDAATLWGQGALGGANVLTF", "cdr3_old": "CASSWTWDAATLWGQGALGGANVLTF", "fixNeeded": false, "good": true, "jCanonical": true, "jFixType": "NoFixNeeded", "jId": "TRBJ2-6*01", "jStart": 19, "vCanonical": true, "vEnd": 4, "vFixType": "NoFixNeeded", "vId": "TRBV5-5*01"} 0
3051 TRA CAAIYESRGSTLGRLYF TRAV13-1*01 TRAJ18*01 HomoSapiens HLA-A*11:01 B2M MHCI AVFDRKSDAK EBNA4 EBV https://www.10xgenomics.com/resources/application-notes/a-new-way-of-exploring-immunity-linking-highly-multiplexed-antigen-recognition-to-immune-repertoire-and-phenotype/# {"frequency": "1/11684", "identification": "dextramer-sort", "sequencing": "rna-seq", "singlecell": "yes", "verification": ""} {"cell.subset": "", "clone.id": "", "donor.MHC": "", "donor.MHC.method": "", "epitope.id": "", "replica.id": "", "samples.found": 1, "structure.id": "", "studies.found": 1, "study.id": "", "subject.cohort": "", "subject.id": "1", "tissue": ""} {"cdr3": "CAAIYESRGSTLGRLYF", "cdr3_old": "CAAIYESRGSTLGRLYF", "fixNeeded": false, "good": true, "jCanonical": true, "jFixType": "NoFixNeeded", "jId": "TRAJ18*01", "jStart": 7, "oldVEnd": -1, "oldVFixType": "FailedBadSegment", "oldVId": null, "vCanonical": true, "vEnd": 3, "vFixType": "ChangeSegment", "vId": "TRAV13-1*01"} 0
15761 TRA CALRLNNQGGKLIF TRAV9-2*01 TRAJ23*01 HomoSapiens HLA-A*03:01 B2M MHCI KLGGALQAK IE1 CMV https://www.10xgenomics.com/resources/application-notes/a-new-way-of-exploring-immunity-linking-highly-multiplexed-antigen-recognition-to-immune-repertoire-and-phenotype/# {"frequency": "1/25584", "identification": "dextramer-sort", "sequencing": "rna-seq", "singlecell": "yes", "verification": ""} {"cell.subset": "", "clone.id": "", "donor.MHC": "", "donor.MHC.method": "", "epitope.id": "", "replica.id": "", "samples.found": 1, "structure.id": "", "studies.found": 1, "study.id": "", "subject.cohort": "", "subject.id": "3", "tissue": ""} {"cdr3": "CALRLNNQGGKLIF", "cdr3_old": "CALRLNNQGGKLIF", "fixNeeded": false, "good": true, "jCanonical": true, "jFixType": "NoFixNeeded", "jId": "TRAJ23*01", "jStart": 6, "vCanonical": true, "vEnd": 3, "vFixType": "NoFixNeeded", "vId": "TRAV9-2*01"} 0
"""
path = PathBuilder.build(EnvironmentSettings.root_path +
"test/tmp/trcdist_encoder/")
with open(path + "receptors.tsv", "w") as file:
file.writelines(file_content)
params = DefaultParamsLoader.load(
EnvironmentSettings.default_params_path + "datasets/", "vdjdb")
params["is_repertoire"] = False
params["paired"] = True
params["result_path"] = path
params["path"] = path
params["sequence_file_size"] = 1
params["receptor_chains"] = "TRA_TRB"
params['organism'] = 'human'
dataset = VDJdbImport.import_dataset(params, "vdjdb_dataset")
encoder = TCRdistEncoder.build_object(dataset, **{"cores": 2})
encoded_dataset = encoder.encode(
dataset,
EncoderParams(f"{path}result/",
LabelConfiguration([Label("epitope")])))
self.assertTrue(encoded_dataset.encoded_data.examples.shape[0]
== encoded_dataset.encoded_data.examples.shape[1]
and encoded_dataset.encoded_data.examples.shape[0]
== dataset.get_example_count())
shutil.rmtree(path)
def test_encode(self):
path = EnvironmentSettings.tmp_test_path + "count_encoder/"
PathBuilder.build(path)
repertoires, metadata = RepertoireBuilder.build(
[["GGG", "III", "LLL", "MMM"],
["DDD", "EEE", "FFF", "III", "LLL", "MMM"], ["CCC", "FFF", "MMM"],
["AAA", "CCC", "EEE", "FFF", "LLL", "MMM"]],
labels={"l1": [True, True, False, False]},
path=path)
dataset = RepertoireDataset(repertoires=repertoires,
metadata_file=metadata,
identifier="1")
encoder = SequenceCountEncoder.build_object(
dataset, **{
"comparison_attributes": ["sequence_aas"],
"p_value_threshold": 0.4,
"sequence_batch_size": 4
})
label_config = LabelConfiguration(
[Label("l1", [True, False], positive_class=True)])
encoded_dataset = encoder.encode(
dataset, EncoderParams(result_path=path,
label_config=label_config))
test = encoded_dataset.encoded_data.examples
self.assertTrue(test[0] == 1)
self.assertTrue(test[1] == 1)
self.assertTrue(test[2] == 0)
self.assertTrue(test[3] == 0)
self.assertTrue("III" in encoded_dataset.encoded_data.feature_names)
shutil.rmtree(path)
def add_label(self,
label: str,
values: list = None,
auxiliary_labels: list = None,
positive_class=None):
vals = list(values) if values else None
if label in self._labels and self._labels[label] is not None and len(
self._labels[label]) > 0:
warnings.warn(
"Label " + label +
" has already been set. Overriding existing values...",
Warning)
if positive_class is not None:
ParameterValidator.assert_in_valid_list(positive_class, values,
Label.__name__,
'positive_class')
self._labels[label] = Label(label, vals, auxiliary_labels,
positive_class)
def test_generate(self):
path = PathBuilder.build(f"{EnvironmentSettings.tmp_test_path}kernel_sequence_logo/")
dataset = RandomDatasetGenerator.generate_receptor_dataset(receptor_count=500, chain_1_length_probabilities={4: 1},
chain_2_length_probabilities={4: 1},
labels={"CMV": {True: 0.5, False: 0.5}}, path=path + "dataset/")
enc_dataset = OneHotReceptorEncoder(True, 1, False, "enc1").encode(dataset, EncoderParams(path + "result/",
LabelConfiguration([Label("CMV", [True, False])])))
cnn = ReceptorCNN(kernel_count=2, kernel_size=[3], positional_channels=3, sequence_type="amino_acid", device="cpu",
number_of_threads=4, random_seed=1, learning_rate=0.01, iteration_count=10, l1_weight_decay=0.1, evaluate_at=5,
batch_size=100, training_percentage=0.8, l2_weight_decay=0.0)
cnn.fit(enc_dataset.encoded_data, "CMV")
report = KernelSequenceLogo(method=cnn, result_path=path + "logos/")
report.generate_report()
self.assertTrue(os.path.isfile(f"{path}logos/alpha_kernel_3_1.png"))
self.assertTrue(os.path.isfile(f"{path}logos/alpha_kernel_3_2.png"))
self.assertTrue(os.path.isfile(f"{path}logos/beta_kernel_3_1.png"))
self.assertTrue(os.path.isfile(f"{path}logos/beta_kernel_3_2.png"))
self.assertTrue(os.path.isfile(f"{path}logos/alpha_kernel_3_1.csv"))
self.assertTrue(os.path.isfile(f"{path}logos/alpha_kernel_3_2.csv"))
self.assertTrue(os.path.isfile(f"{path}logos/beta_kernel_3_1.csv"))
self.assertTrue(os.path.isfile(f"{path}logos/beta_kernel_3_2.csv"))
self.assertTrue(os.path.isfile(f"{path}logos/fully_connected_layer_weights.csv"))
self.assertTrue(os.path.isfile(f"{path}logos/fully_connected_layer_weights.html"))
shutil.rmtree(path)
def test_fit(self):
path = PathBuilder.build(EnvironmentSettings.tmp_test_path +
"kmermil/")
repertoire_count = 10
dataset = RandomDatasetGenerator.generate_repertoire_dataset(
repertoire_count=repertoire_count,
sequence_count_probabilities={2: 1},
sequence_length_probabilities={4: 1},
labels={"l1": {
True: 0.5,
False: 0.5
}},
path=path + "dataset/")
enc_dataset = AtchleyKmerEncoder(
2, 1, 1, 'relative_abundance', False).encode(
dataset,
EncoderParams(path + "result/",
LabelConfiguration([Label("l1",
[True, False])])))
cls = AtchleyKmerMILClassifier(iteration_count=10,
threshold=-0.0001,
evaluate_at=2,
use_early_stopping=False,
random_seed=1,
learning_rate=0.01,
zero_abundance_weight_init=True,
number_of_threads=8)
cls.fit(enc_dataset.encoded_data, "l1")
predictions = cls.predict(enc_dataset.encoded_data, "l1")
self.assertEqual(repertoire_count, len(predictions["l1"]))
self.assertEqual(
repertoire_count,
len([pred for pred in predictions["l1"]
if isinstance(pred, bool)]))
predictions_proba = cls.predict_proba(enc_dataset.encoded_data, "l1")
self.assertEqual(repertoire_count,
np.rint(np.sum(predictions_proba["l1"])))
self.assertEqual(repertoire_count, predictions_proba["l1"].shape[0])
cls.store(path + "model_storage/",
feature_names=enc_dataset.encoded_data.feature_names)
cls2 = AtchleyKmerMILClassifier(iteration_count=10,
threshold=-0.0001,
evaluate_at=2,
use_early_stopping=False,
random_seed=1,
learning_rate=0.01,
zero_abundance_weight_init=True,
number_of_threads=8)
cls2.load(path + "model_storage/")
cls2_vars = vars(cls2)
del cls2_vars["logistic_regression"]
cls_vars = vars(cls)
del cls_vars["logistic_regression"]
for item, value in cls_vars.items():
if not isinstance(value, np.ndarray):
loaded_value = cls2_vars[item]
self.assertEqual(value, loaded_value)
model = cls.get_model("l1")
self.assertEqual(vars(cls), model)
shutil.rmtree(path)
def test_find_label_associated_sequence_p_values(self):
path = EnvironmentSettings.tmp_test_path + "comparison_data_find_label_assocseqpvalues/"
PathBuilder.build(path)
repertoires = [Repertoire.build_from_sequence_objects([ReceptorSequence()], path, {
"l1": val, "subject_id": subject_id
}) for val, subject_id in zip([True, True, False, False], ["rep_0", "rep_1", "rep_2", "rep_3"])]
col_name_index = {repertoires[index].identifier: index for index in range(len(repertoires))}
comparison_data = ComparisonData(repertoire_ids=[repertoire.identifier for repertoire in repertoires],
comparison_attributes=["sequence_aas"], sequence_batch_size=4, path=path)
comparison_data.batches = [ComparisonDataBatch(**{'matrix': np.array([[1., 0., 0., 0.],
[1., 1., 0., 0.]]),
'items': [('GGG',), ('III',)],
'repertoire_index_mapping': col_name_index, 'path': path, 'identifier': 0}),
ComparisonDataBatch(**{'matrix': np.array([[1., 1., 0., 1.],
[1., 1., 1., 1.]]),
'items': [('LLL',), ('MMM',)],
'repertoire_index_mapping': col_name_index, 'path': path, 'identifier': 1}),
ComparisonDataBatch(**{'matrix': np.array([[0., 1., 0., 0.],
[0., 1., 0., 1.]]),
'items': [('DDD',), ('EEE',)],
'repertoire_index_mapping': col_name_index, 'path': path, 'identifier': 2}),
ComparisonDataBatch(**{'matrix': np.array([[0., 1., 1., 1.],
[0., 0., 1., 1.]]),
'items': [('FFF',), ('CCC',)],
'repertoire_index_mapping': col_name_index, 'path': path, 'identifier': 3}),
ComparisonDataBatch(**{'matrix': np.array([[0., 0., 0., 1.]]),
'items': [('AAA',)],
'repertoire_index_mapping': col_name_index, 'path': path, 'identifier': 4})]
p_values = SequenceFilterHelper.find_label_associated_sequence_p_values(comparison_data, repertoires, Label('l1', [True, False], positive_class=True))
print(p_values)
self.assertTrue(
np.allclose([SequenceFilterHelper.INVALID_P_VALUE, 0.1666666666666667, 0.5000000000000001, 1., SequenceFilterHelper.INVALID_P_VALUE,
0.8333333333333331, 1., 1., 2], p_values, equal_nan=True))
shutil.rmtree(path)
def test_encode(self):
path = PathBuilder.build(EnvironmentSettings.tmp_test_path + "atchley_kmer_encoding/")
dataset = RandomDatasetGenerator.generate_repertoire_dataset(3, {1: 1}, {4: 1}, {"l1": {True: 0.4, False: 0.6}}, path + "dataset/")
encoder = AtchleyKmerEncoder.build_object(dataset, **{"k": 2, "skip_first_n_aa": 1, "skip_last_n_aa": 1, "abundance": "RELATIVE_ABUNDANCE",
"normalize_all_features": False})
encoded_dataset = encoder.encode(dataset, EncoderParams(path + "result/", LabelConfiguration(labels=[Label("l1")])))
self.assertEqual((3, 11, 3), encoded_dataset.encoded_data.examples.shape)
self.assertEqual(0., encoded_dataset.encoded_data.examples[0, -1, 0])
shutil.rmtree(path)
def test_run(self):
path = EnvironmentSettings.tmp_test_path + "hpoptimproc/"
PathBuilder.build(path)
repertoires, metadata = RepertoireBuilder.build(
sequences=[["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"],
["AAA", "CCC", "DDD"], ["AAA", "CCC", "DDD"]],
path=path,
labels={
"l1": [
1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2,
1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2
],
"l2": [
0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 1, 0, 1,
0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1
]
})
dataset = RepertoireDataset(repertoires=repertoires,
metadata_file=metadata,
params={
"l1": [1, 2],
"l2": [0, 1]
})
enc1 = {
"k": 3,
"model_type": ModelType.SEQUENCE.name,
"vector_size": 4
}
enc2 = {
"k": 3,
"model_type": ModelType.SEQUENCE.name,
"vector_size": 6
}
hp_settings = [
HPSetting(Word2VecEncoder.build_object(dataset, **enc1), enc1,
LogisticRegression(), {
"model_selection_cv": False,
"model_selection_n_folds": -1
}, []),
HPSetting(
Word2VecEncoder.build_object(dataset, **enc2), enc2, SVM(), {
"model_selection_cv": False,
"model_selection_n_folds": -1
},
[ClonesPerRepertoireFilter(lower_limit=-1, upper_limit=1000)])
]
report = SequenceLengthDistribution()
label_config = LabelConfiguration(
[Label("l1", [1, 2]), Label("l2", [0, 1])])
process = TrainMLModelInstruction(
dataset, GridSearch(hp_settings), hp_settings,
SplitConfig(SplitType.RANDOM,
1,
0.5,
reports=ReportConfig(data_splits={"seqlen": report})),
SplitConfig(SplitType.RANDOM,
1,
0.5,
reports=ReportConfig(data_splits={"seqlen": report})),
{Metric.BALANCED_ACCURACY}, Metric.BALANCED_ACCURACY, label_config,
path)
state = process.run(result_path=path)
self.assertTrue(isinstance(state, TrainMLModelState))
self.assertEqual(1, len(state.assessment_states))
self.assertTrue("l1" in state.assessment_states[0].label_states)
self.assertTrue("l2" in state.assessment_states[0].label_states)
shutil.rmtree(path)
def test_generate(self):
path = EnvironmentSettings.tmp_test_path + "cv_feature_performance/"
state = TrainMLModelState(
assessment=SplitConfig(split_count=5,
split_strategy=SplitType.K_FOLD),
selection=SplitConfig(split_count=10,
split_strategy=SplitType.K_FOLD),
optimization_metric=Metric.ACCURACY,
label_configuration=LabelConfiguration(
labels=[Label(name="CMV", values=[True, False])]),
hp_settings=[
HPSetting(encoder_params={"p_value_threshold": 0.001},
encoder_name="e1",
encoder=SequenceAbundanceEncoder([], 0, 0, 0),
preproc_sequence=[],
ml_method_name="ml1",
ml_method=ProbabilisticBinaryClassifier(10, 0.1),
ml_params={}),
HPSetting(encoder_params={"p_value_threshold": 0.01},
encoder_name="e2",
encoder=SequenceAbundanceEncoder([], 0, 0, 0),
preproc_sequence=[],
ml_method_name="ml1",
ml_method=ProbabilisticBinaryClassifier(10, 0.1),
ml_params={}),
HPSetting(encoder_params={"p_value_threshold": 0.01},
encoder=SequenceAbundanceEncoder([], 0, 0, 0),
preproc_sequence=[],
ml_method=ProbabilisticBinaryClassifier(10, 0.01),
ml_params={})
],
dataset=None,
hp_strategy=None,
metrics=None)
report = CVFeaturePerformance("p_value_threshold",
state,
path,
is_feature_axis_categorical=True,
name="report1")
with self.assertWarns(RuntimeWarning):
report.generate_report()
state.hp_settings = state.hp_settings[:2]
state.assessment_states = [
HPAssessmentState(i, None, None, None, state.label_configuration)
for i in range(state.assessment.split_count)
]
for assessment_state in state.assessment_states:
assessment_state.label_states["CMV"] = HPLabelState("CMV", [])
assessment_state.label_states["CMV"].assessment_items = {
setting.get_key():
HPItem(performance={'accuracy': random.uniform(0.5, 1)},
hp_setting=setting)
for setting in state.hp_settings
}
assessment_state.label_states[
"CMV"].selection_state = HPSelectionState(
[], [], "", GridSearch(state.hp_settings))
assessment_state.label_states["CMV"].selection_state.hp_items = {
str(setting): [
HPItem(performance={'accuracy': random.uniform(0.5, 1)},
hp_setting=setting)
for _ in range(state.selection.split_count)
]
for setting in state.hp_settings
}
report.state = state
report_result = report.generate_report()
self.assertTrue(isinstance(report_result, ReportResult))
self.assertEqual(2, len(report_result.output_tables))
self.assertEqual(1, len(report_result.output_figures))
self.assertTrue(os.path.isfile(report_result.output_figures[0].path))
self.assertTrue(os.path.isfile(report_result.output_tables[0].path))
self.assertTrue(os.path.isfile(report_result.output_tables[1].path))
shutil.rmtree(path)
def test_fit(self):
path = PathBuilder.build(EnvironmentSettings.tmp_test_path + "cnn/")
dataset = RandomDatasetGenerator.generate_receptor_dataset(
receptor_count=500,
chain_1_length_probabilities={4: 1},
chain_2_length_probabilities={4: 1},
labels={"CMV": {
True: 0.5,
False: 0.5
}},
path=path + "dataset/")
enc_dataset = OneHotReceptorEncoder(True, 1, False, "enc1").encode(
dataset,
EncoderParams(path + "result/",
LabelConfiguration([Label("CMV", [True, False])])))
cnn = ReceptorCNN(kernel_count=2,
kernel_size=[3],
positional_channels=3,
sequence_type="amino_acid",
device="cpu",
number_of_threads=4,
random_seed=1,
learning_rate=0.01,
iteration_count=10,
l1_weight_decay=0.1,
evaluate_at=5,
batch_size=100,
training_percentage=0.8,
l2_weight_decay=0.0)
cnn.fit(encoded_data=enc_dataset.encoded_data, label_name="CMV")
predictions = cnn.predict(enc_dataset.encoded_data, "CMV")
self.assertEqual(500, len(predictions["CMV"]))
self.assertEqual(
500,
len([
pred for pred in predictions["CMV"] if isinstance(pred, bool)
]))
predictions_proba = cnn.predict_proba(enc_dataset.encoded_data, "CMV")
self.assertEqual(500, np.rint(np.sum(predictions_proba["CMV"])))
self.assertEqual(500, predictions_proba["CMV"].shape[0])
cnn.store(path + "model_storage/")
cnn2 = ReceptorCNN(sequence_type="amino_acid")
cnn2.load(path + "model_storage/")
cnn2_vars = vars(cnn2)
del cnn2_vars["CNN"]
cnn_vars = vars(cnn)
del cnn_vars["CNN"]
for item, value in cnn_vars.items():
if not isinstance(value, np.ndarray):
self.assertEqual(value, cnn2_vars[item])
model = cnn.get_model(["CMV"])
self.assertEqual(vars(cnn), model)
shutil.rmtree(path)