working on liberating the titration script from the experiments directory

experiments/dataset-size-sensitivity.py

@@ -21,12 +21,13 @@ Plot AUC and F1 score of predictors as a function of dataset size
 from argparse import ArgumentParser
 
 import numpy as np
-import mhcflurry
 import sklearn
 import sklearn.metrics
-from sklearn.linear_model import LinearRegression
 import seaborn
 
+from mhcflurry.dataset import Dataset
+from mhcflurry.class1_binding_predictor import Class1BindingPredictor
+
 from dataset_paths import PETERS2009_CSV_PATH
 
 parser = ArgumentParser()
@@ -42,7 +43,7 @@ parser.add_argument(
 parser.add_argument(
     "--max-ic50",
     type=float,
-    default=20000.0)
+    default=50000.0)
 
 parser.add_argument(
     "--hidden-layer-size",
@@ -50,6 +51,12 @@ parser.add_argument(
     default=10,
     help="Hidden layer size for neural network, if 0 use linear regression")
 
+parser.add_argument(
+    "--embedding-dim",
+    type=int,
+    default=50,
+    help="Number of dimensions for vector embedding of amino acids")
+
 parser.add_argument(
     "--activation",
     default="tanh")
@@ -71,35 +78,41 @@ parser.add_argument(
     help="How many times to train model for same dataset size")
 
 
-def binary_encode(X, n_indices=20):
-    n_cols = X.shape[1]
-    X_encode = np.zeros((len(X), n_indices * n_cols), dtype=float)
-    for i in range(len(X)):
-        for col_idx in range(n_cols):
-            X_encode[i, col_idx * n_indices + X[i, col_idx]] = True
-    return X_encode
-
-
 def subsample_performance(
-        X,
-        Y,
-        max_ic50,
+        dataset,
+        allele,
+        pretraining=False,
         model_fn=None,
-        fractions=np.arange(0.01, 1, 0.03),
+        n_training_samples=[
+            25,
+            50,
+            100,
+            200,
+            300,
+            400,
+            500,
+            600,
+            800,
+            1000,
+            1200,
+            1400,
+            1600,
+            1800,
+            2000],
         niters=10,
-        fraction_test=0.2,
-        nb_epoch=50,
+        n_training_epochs=200,
         batch_size=32):
-    n = len(Y)
+    n_total = len(dataset)
+
     xs = []
     aucs = []
     f1s = []
+
     for iternum in range(niters):
-        if model_fn is None:
-            model = LinearRegression()
-        else:
-            model = model_fn()
-            initial_weights = model.get_weights()
+
+        model = model_fn()
+        initial_weights = model.get_weights()
+
         mask = np.random.rand(n) > fraction_test
         X_train = X[mask]
         X_test = X[~mask]
@@ -139,22 +152,21 @@ def subsample_performance(
 
 if __name__ == "__main__":
     args = parser.parse_args()
-    print(args)
-    datasets, _ = mhcflurry.data_helpers.load_data(
-        args.training_csv,
-        binary_encoding=True,
-        flatten_binary_encoding=True,
-        max_ic50=args.max_ic50)
-    dataset = datasets[args.allele]
+    dataset = Dataset.load_csv(args.training_csv)
     X = dataset.X
     Y = dataset.Y
     print("Total # of samples for %s: %d" % (args.allele, len(Y)))
-    if args.hidden_layer_size > 0:
-        model_fn = lambda: mhcflurry.feedforward.make_hotshot_network(
-            layer_sizes=[args.hidden_layer_size],
-            activation=args.activation)
+    if args.hidden_layer_size == 0:
+        hidden_layer_sizes = []
     else:
-        model_fn = None
+        hidden_layer_sizes = [args.hidden_layer_size]
+
+    def make_model():
+        return Class1BindingPredictor.from_hyperparameters(
+            layer_sizes=[args.hidden_layer_size],
+            activation=args.activation,
+            embedding_dim=args.embedding_dim)
+
     xs, aucs, f1s = subsample_performance(
         X=X,
         Y=Y,

mhcflurry/class1_allele_specific_kmer_ic50_predictor_base.py

@@ -43,10 +43,11 @@ class Class1AlleleSpecificKmerIC50PredictorBase(IC50PredictorBase):
             verbose,
             max_ic50=MAX_IC50,
             kmer_size=9):
-        self.name = name
-        self.max_ic50 = max_ic50
+        IC50PredictorBase.__init__(
+            name=name,
+            max_ic50=max_ic50,
+            verbose=verbose)
         self.allow_unknown_amino_acids = allow_unknown_amino_acids
-        self.verbose = verbose
         self.kmer_size = kmer_size
 
     def __repr__(self):

mhcflurry/dataset.py

@@ -254,7 +254,9 @@ class Dataset(object):
             if len(column) != len(alleles):
                 raise ValueError(
                     "Wrong length for column '%s', expected %d but got %d" % (
-                        column_name, column))
+                        column_name,
+                        len(alleles),
+                        len(column)))
             df[column_name] = np.asarray(column)
         return cls(df)
 

mhcflurry/ic50_predictor_base.py

@@ -32,7 +32,7 @@ class IC50PredictorBase(object):
     def __init__(
             self,
             name,
-            verbose,
+            verbose=False,
             max_ic50=MAX_IC50):
         self.name = name
         self.max_ic50 = max_ic50
@@ -42,7 +42,6 @@ class IC50PredictorBase(object):
         return "%s(name=%s, max_ic50=%f)" % (
             self.__class__.__name__,
             self.name,
-            self.model,
             self.max_ic50)
 
     def __str__(self):

script/mhcflurry-train-class1-allele-specific-models.py

@@ -167,6 +167,13 @@ if __name__ == "__main__":
             print("-- too few data points, skipping")
             continue
 
+        model.fit_dataset(
+            allele_dataset,
+            pretraining_dataset=imputed_dataset_allele,
+            n_training_epochs=args.training_epochs,
+            n_random_negative_samples=args.random_negative_samples,
+            verbose=True)
+
         if exists(json_path):
             print("-- removing old model description %s" % json_path)
             remove(json_path)
@@ -175,13 +182,6 @@ if __name__ == "__main__":
             print("-- removing old weights file %s" % hdf_path)
             remove(hdf_path)
 
-        model.fit_dataset(
-            allele_dataset,
-            pretraining_dataset=imputed_dataset_allele,
-            n_training_epochs=args.training_epochs,
-            n_random_negative_samples=args.random_negative_samples,
-            verbose=True)
-
         model.to_disk(
             model_json_path=json_path,
             weights_hdf_path=hdf_path,