Hope anyone can help a beginner here. I'm building a proof-of concept tensorflow classifier for DNA sequences. However, the NN model does not let through train and test vectors saying the matrix size is incompatible. Where could be the mistake in my implementation?

import numpy as np
import pandas as pd

df=pd.DataFrame(columns=['sequences', 'label'])
a='ttttccagaattctcttagttt gtgatgtctttattgcttctattt'
b='ctcctgcttgctttttttcttg ggtttctgatattctttaaaggat'
c='tcctgcttgctttttttcttgg gtttctgatattctttaaaggatt'

from sklearn.preprocessing import LabelEncoder, OneHotEncoder
integer_encoder = LabelEncoder()
one_hot_encoder = OneHotEncoder(categories='auto')
input_features = []
for sequence in sequences:
     integer_encoded = integer_encoder.fit_transform(list(sequence))
     integer_encoded = np.array(integer_encoded).reshape(-1, 1)
     one_hot_encoded = one_hot_encoder.fit_transform(integer_encoded)

input_features = np.stack(input_features)
print("Example sequence\n-----------------------")
print('DNA Sequence #1:\n',sequences[0][:10],'...',sequences[0][-10:])
print('One hot encoding of Sequence #1:\n',input_features[0].T)

one_hot_encoder = OneHotEncoder(categories='auto')
labels = np.array(labels).reshape(-1, 1)
input_labels = one_hot_encoder.fit_transform(labels).toarray()
print('One-hot encoded labels:\n',input_labels.T)

from sklearn.model_selection import train_test_split
train_features, test_features, train_labels, test_labels = train_test_split(input_features, input_labels, test_size=0.25, random_state=42)

def get_batch(x_data, y_data, batch_size):
     idxs = np.random.randint(0, len(y_data), batch_size)
     return x_data[idxs,:,:], y_data[idxs]

epochs = 10
batch_size = 100

x_train = train_features
x_test = train_labels
x_train = x_train / 255.0
x_test = x_test / 255.0
x_test = tf.Variable(x_test)

W1 = tf.Variable(tf.random.normal([784, 300], stddev=0.03), name='W1')
b1 = tf.Variable(tf.random.normal([300]), name='b1')
W2 = tf.Variable(tf.random.normal([300, 10], stddev=0.03), name='W2')
b2 = tf.Variable(tf.random.normal([10]), name='b2')

def nn_model(x_input, W1, b1, W2, b2): 
     x_input = tf.reshape(x_input, (x_input.shape[0], -1))
     x = tf.add(tf.matmul(tf.cast(x_input, tf.float32), W1), b1) #ERROR FROM HERE?
     x = tf.nn.relu(x)
     logits = tf.add(tf.matmul(x, W2), b2)
     return logits

def loss_fn(logits, labels):
     cross_entropy = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=labels,logits=logits))
     return cross_entropy

optimizer = tf.keras.optimizers.Adam()
total_batch = int(len(y_train) / batch_size)

for epoch in range(epochs):
    avg_loss = 0
    for i in range(total_batch):
        batch_x, batch_y = get_batch(x_train, y_train, batch_size=batch_size)
        # create tensors
        batch_x = tf.Variable(batch_x)
        batch_y = tf.Variable(batch_y)
        # create a one hot vector
        batch_y = tf.one_hot(batch_y, 10)
        with tf.GradientTape() as tape:
            logits = nn_model(batch_x, W1, b1, W2, b2)
            loss = loss_fn(logits, batch_y)
        gradients = tape.gradient(loss, [W1, b1, W2, b2])
        optimizer.apply_gradients(zip(gradients, [W1, b1, W2, b2]))
        avg_loss += loss / total_batch
    test_logits = nn_model(x_test, W1, b1, W2, b2) ##ERROR LOG APPEARS HERE
    max_idxs = tf.argmax(test_logits, axis=1)
    test_acc = np.sum(max_idxs.numpy() == y_test) / len(y_test)
    print(f"Epoch: {epoch + 1}, loss={avg_loss:.3f}, test set accuracy = {test_acc*100:.3f}%")
print("\nTraining complete!")

Traceback (most recent call last): File "", line 16, in File "", line 4, in nn_model File "/usr/local/lib/python3.8/dist-packages/tensorflow/python/util/traceback_utils.py", line 153, in error_handler raise e.with_traceback(filtered_tb) from None File "/usr/local/lib/python3.8/dist-packages/tensorflow/python/framework/ops.py", line 7107, in raise_from_not_ok_status raise core._status_to_exception(e) from None # pylint: disable=protected-access tensorflow.python.framework.errors_impl.InvalidArgumentError: Matrix size-incompatible: In[0]: [2,1], In[1]: [784,300] [Op:MatMul]

  • $\begingroup$ I would check and make sure that the dimensions of your matrix multiplication make sense. Specifically, figure out if you can reproduce the error by just multiplying matrices of the size you're using, since as you point out that seems to be where the error comes from. $\endgroup$ Feb 24, 2023 at 6:43

1 Answer 1


The sci-kit learn looks fine, for a proof of concept. Furthermore, the deep learning looks good, e.g. Adams, loss etc .. all good. Of course it at ain't RNN or CNN but its proof of concept.

However, the error you encountered suggests it is about resizing the numpy array and that is better answered on Stackoverflow in the absence of an answer here. Dealing with pandas is one thing, munging a numpy array is pretty hardcore. Its a single line of code, but you've got to know it exactly.

An example is where I d focus,

tf.reshape(x_input, (x_input.shape[0], -1))

Before or whether you need to post there please check:

  1. Where is import tensorflow as tf ?
  2. Using keras without e.g. from tensorflow.keras import optimizers (well okay you can do that given point 1 but generally the keras method is specifically imported)

If I haven't misread the code it will fall over, as soon as it gets past sci-kit learn. The advantage of placing all imports at the top of the code is that its easy to read whats going on. There is another issue, but if a big deal Stackoverflow (or someone else here) will flag it.

Just to we explain point 2, I suspect the tensor flow on question is Tensorflow 2 if this is true then Keras is built in and thus tf.keras is fine

  • $\begingroup$ Thank you for the answer. import tensorflow as tf ; tf.reshape(x_input, (x_input.shape[0], -1)) does not help to solve the error... $\endgroup$
    – monade
    Jan 31, 2022 at 12:29
  • $\begingroup$ Okay fine, just my guess about a point in the code that could generate the error. $\endgroup$
    – M__
    Jan 31, 2022 at 13:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.