From 39603f73d1d885789877f2def1f11f7ab9e0db10 Mon Sep 17 00:00:00 2001
From: Tim O'Donnell <timodonnell@gmail.com>
Date: Mon, 15 Jul 2019 11:55:12 -0400
Subject: [PATCH] Add expensive_test_pretrain_optimizable.py
---
mhcflurry/train_pan_allele_models_command.py | 2 -
test/expensive_test_pretrain_optimizable.py | 92 ++++++++++++++++++++
test/test_train_pan_allele_models_command.py | 2 +-
3 files changed, 93 insertions(+), 3 deletions(-)
create mode 100644 test/expensive_test_pretrain_optimizable.py
diff --git a/mhcflurry/train_pan_allele_models_command.py b/mhcflurry/train_pan_allele_models_command.py
index 95264ca3..1fcc9c04 100644
--- a/mhcflurry/train_pan_allele_models_command.py
+++ b/mhcflurry/train_pan_allele_models_command.py
@@ -290,7 +290,6 @@ def main(args):
GLOBAL_DATA["train_data"] = df
GLOBAL_DATA["folds_df"] = folds_df
GLOBAL_DATA["allele_encoding"] = allele_encoding
- GLOBAL_DATA["args"] = args
if not os.path.exists(args.out_models_dir):
print("Attempting to create directory: %s" % args.out_models_dir)
@@ -444,7 +443,6 @@ def train_model(
df = constant_data["train_data"]
folds_df = constant_data["folds_df"]
allele_encoding = constant_data["allele_encoding"]
- args = constant_data["args"]
if predictor is None:
predictor = Class1AffinityPredictor(
diff --git a/test/expensive_test_pretrain_optimizable.py b/test/expensive_test_pretrain_optimizable.py
new file mode 100644
index 00000000..59cd8105
--- /dev/null
+++ b/test/expensive_test_pretrain_optimizable.py
@@ -0,0 +1,92 @@
+# Expensive test - not run by default.
+
+from mhcflurry import train_pan_allele_models_command
+from mhcflurry.downloads import get_path
+from mhcflurry.allele_encoding import AlleleEncoding
+
+import pandas
+import numpy
+
+PRETRAIN_DATA_PATH = get_path(
+ "random_peptide_predictions", "predictions.csv.bz2")
+
+FULL_TRAIN_DF = pandas.read_csv(
+ get_path(
+ "data_curated",
+ "curated_training_data.no_mass_spec.csv.bz2"))
+TRAIN_DF = FULL_TRAIN_DF.loc[
+ (FULL_TRAIN_DF.peptide.str.len() >= 8) &
+ (FULL_TRAIN_DF.peptide.str.len() <= 15)
+]
+ALLELE_SEQUENCES = pandas.read_csv(
+ get_path("allele_sequences", "allele_sequences.csv"),
+ index_col=0).sequence
+ALLELE_SEQUENCES = ALLELE_SEQUENCES.loc[
+ ALLELE_SEQUENCES.index.isin(TRAIN_DF.allele)
+]
+TRAIN_DF = TRAIN_DF.loc[
+ TRAIN_DF.allele.isin(ALLELE_SEQUENCES.index)
+]
+FOLDS_DF = pandas.DataFrame(index=TRAIN_DF.index)
+FOLDS_DF["fold_0"] = True
+
+HYPERPARAMTERS = {
+ 'activation': 'tanh', 'allele_dense_layer_sizes': [],
+ 'batch_normalization': False,
+ 'dense_layer_l1_regularization': 9.999999999999999e-11,
+ 'dense_layer_l2_regularization': 0.0, 'dropout_probability': 0.5,
+ 'early_stopping': True, 'init': 'glorot_uniform',
+ 'layer_sizes': [1024, 512], 'learning_rate': None,
+ 'locally_connected_layers': [], 'loss': 'custom:mse_with_inequalities',
+ 'max_epochs': 1, 'min_delta': 0.0, 'minibatch_size': 128,
+ 'optimizer': 'rmsprop', 'output_activation': 'sigmoid', 'patience': 20,
+ 'peptide_allele_merge_activation': '',
+ 'peptide_allele_merge_method': 'concatenate',
+ 'peptide_amino_acid_encoding': 'BLOSUM62', 'peptide_dense_layer_sizes': [],
+ 'peptide_encoding': {'alignment_method': 'left_pad_centered_right_pad',
+ 'max_length': 15, 'vector_encoding_name': 'BLOSUM62'},
+ 'random_negative_affinity_max': 50000.0,
+ 'random_negative_affinity_min': 20000.0, 'random_negative_constant': 25,
+ 'random_negative_distribution_smoothing': 0.0,
+ 'random_negative_match_distribution': True, 'random_negative_rate': 0.2,
+ 'train_data': {'pretrain': True,
+ 'pretrain_max_epochs': 1,
+ 'pretrain_peptides_per_epoch': 1024,
+ 'pretrain_steps_per_epoch': 16},
+ 'validation_split': 0.1,
+}
+
+
+def test_optimizable():
+ predictor = train_pan_allele_models_command.train_model(
+ work_item_num=0,
+ num_work_items=1,
+ architecture_num=0,
+ num_architectures=1,
+ fold_num=0,
+ num_folds=1,
+ replicate_num=0,
+ num_replicates=1,
+ hyperparameters=HYPERPARAMTERS,
+ pretrain_data_filename=PRETRAIN_DATA_PATH,
+ verbose=1,
+ progress_print_interval=5.0,
+ predictor=None,
+ save_to=None,
+ constant_data={
+ 'train_data': TRAIN_DF,
+ 'folds_df': FOLDS_DF,
+ 'allele_encoding': AlleleEncoding(
+ alleles=ALLELE_SEQUENCES.index.values,
+ allele_to_sequence=ALLELE_SEQUENCES.to_dict()),
+ },
+ )
+ (network,) = predictor.neural_networks
+ pretrain_val_loss = network.fit_info[-1]['training_info']["val_loss"][-1]
+ print(predictor)
+ print(pretrain_val_loss)
+ numpy.testing.assert_array_less(pretrain_val_loss, 0.1)
+
+
+if __name__ == "__main__":
+ test_optimizable()
diff --git a/test/test_train_pan_allele_models_command.py b/test/test_train_pan_allele_models_command.py
index 52c747b0..9fafea45 100644
--- a/test/test_train_pan_allele_models_command.py
+++ b/test/test_train_pan_allele_models_command.py
@@ -95,7 +95,7 @@ HYPERPARAMETERS_LIST = [
"pretrain": True,
'pretrain_peptides_per_epoch': 128,
'pretrain_max_epochs': 2,
- 'pretrain_max_val_loss': 0.1,
+ 'pretrain_max_val_loss': 0.2,
},
'validation_split': 0.1,
},
--
GitLab