Move hardcorded data to a constants module

src/constants.py

@@ -0,0 +1,3 @@
+BASES = "ACGT"
+TRAIN_DATASET = "data/train_data.tfrecords"
+TEST_DATASET = "data/test_data.tfrecords"

src/model.py

@@ -1,8 +1,10 @@
 from tensorflow.keras import Model, Sequential, layers
 from tensorflow.keras.regularizers import l2
 
+from constants import BASES
 
-def build_model(hyper_parameters, bases="ACGT") -> Model:
+
+def build_model(hyper_parameters) -> Model:
     """
     Builds the CNN model
     """
@@ -39,6 +41,6 @@ def build_model(hyper_parameters, bases="ACGT") -> Model:
             ),
             layers.Dropout(rate=0.3),
             # Output layer with softmax activation
-            layers.Dense(units=len(bases), activation="softmax"),
+            layers.Dense(units=len(BASES), activation="softmax"),
         ]
     )

src/preprocessing.py

@@ -2,9 +2,12 @@ from Bio.motifs import create
 from Bio.SeqIO import parse
 from numpy.random import random
 from tensorflow.io import TFRecordWriter
+from tensorflow.data import TFRecordDataset
 from tensorflow.train import BytesList, Example, Feature, Features, FloatList
 from typing import List
 
+from constants import TRAIN_DATASET, TEST_DATASET
+
 
 def generate_example(sequence, weight_matrix) -> bytes:
     """
@@ -41,14 +44,16 @@ def create_dataset(filepath) -> None:
     """
     data = parse_data(filepath)
     train_test_split = 0.7
-    with TFRecordWriter("data/train_data.tfrecords") as train_writer, TFRecordWriter(
+    with TFRecordWriter(TRAIN_DATASET) as train, TFRecordWriter(TEST_DATASET) as test:
-        "data/test_data.tfrecords"
-    ) as test_writer:
         for element in data:
             if random() < train_test_split:
-                train_writer.write(element)
+                train.write(element)
             else:
-                test_writer.write(element)
+                test.write(element)
+
+
+def read_dataset():
+    pass
 
 
 create_dataset("data/curesim-HVR.fastq")