Deep Learning for the Life Sciences

Applying Deep Learning to Genomics, Microscopy, Drug Discovery & More.

Using DeepChem with TensorFlow 2.x + Keras or PyTorch 🔥, projects and exercises adapted to run on Google Colab.

Repository of the book

Projects and exercises

Chapter	Name	TensorFlow	PyTorch
3 - ML with DeepChem	Predict toxicity of molecules	✔️	✖️
3 - ML with DeepChem	Digit Recognition (MNIST)	✔️	✖️
4 - Molecules	Predict solubility of molecules	✔️	✖️
4 - Molecules	SMARTS Strings	✔️	✖️
5 - Biophysics	Predict affinity of protein-ligands	✔️	✖️
6 - Genomics	Predict TF binding (JUND)	✔️	✖️
6 - Genomics	Predict TF binding with chromatin accessibility	✔️	✖️
6 - Genomics	Predict RNA Interference	✔️	✖️
7 - Microscopy	Cell counting	✔️	✖️
7 - Microscopy	Cell segmentation	✔️	✖️
8 - Medicine	Predict diabetic retinopathy progression	✔️	✖️
9 - Generative models	Generate molecules using MUV dataset	✔️	✖️
10 - Interpretation		-	-
11 - Virtual Screening	Virtual screening work-flow	✔️	-

Notes:

1. On the Jupyter Notebooks the models don't save after/during the training. You must set model_dir and call model.restore() for use trained models, you can read more here.
2. If you wish to run the models to train, please enable GPU on Colab (Runtime > Change runtime type > Hardware accelerator -> GPU).
3. The projects are demonstrative, you shouldn't use them in a real life application, but you can have like reference to create robust models to research, exploration and more.