This repository contains Jupyter notebooks and notes for deep learning for genomics. We principally use keras and scikit-learn for deep learning, but other libraries (joblib, ray, tune, hyperas, hyperopt) are used along the way as well.

• Data Exploration for DNA transcription binding site data set:
  • Jupyter notebook: data_exploration_dna_transcription.ipynb
  • Google CoLab notebook: data_exploration_dna_transcription.ipynb

• Keras 1D CNN for predicting DNA transcription binding sites:

  • Jupyter notebook: keras_cnn1d_dna_transcription.ipynb
  • Google CoLab notebook: keras_cnn1d_dna_transcription.ipynb

• Keras + Scikit-learn Cross-Validated 1D CNN (transcription factor bind site predictions):

  • Jupyter notebook: keras_sklearn_cnn1d_dna_transcription.ipynb
  • Google CoLab notebook: keras_sklearn_cnn1d_dna_transcription.ipynb

• Keras + Scikit-learn Cross-Validated 1D CNN, Applying Log Transform to Chromatin Data:

  • Jupyter notebook: keras_sklearn_cnn1d_dna_transcription_logx.ipynb
  • Google CoLab notebook: keras_sklearn_cnn1d_dna_transcription_logx.ipynb

• Keras + Scikit-learn Cross-Validated 1D CNN, Applying Power Transform to Chromatin Data:

  • Jupyter notebook: keras_sklearn_cnn1d_dna_transcription_powerx.ipynb
  • Google CoLab notebook: keras_sklearn_cnn1d_dna_transcription_powerx.ipynb

• Keras + Scikit-learn Cross-Validated 1D CNN, Applying Quantile Transform to Chromatin Data:

  • Jupyter notebook: keras_sklearn_cnn1d_dna_transcription_quantx.ipynb
  • Google CoLab notebook: keras_sklearn_cnn1d_dna_transcription_quantx.ipynb

• Hyperas for Hyperparameter Optimization of Cross-Validated 1D CNN (transcription factor bind site predictions):

  • Jupyter notebook: hyperas_cnn1d_dna_transcription.ipynb
  • Google CoLab notebook: hyperas_cnn1d_dna_transcription.ipynb

• Ray/Tune for Hyperparameter Optimization of Cross-Validated 1D CNN (IN PROGRESS):

  • Jupyter notebook: ray_tune_cnn1d_dna_transcription.ipynb
  • Google CoLab notebook: ray_tune_cnn1d_dna_transcription.ipynb