8 Introduction to Deep Learning

 

This chapter covers

  • The kinase protein family from perspectives of both ligand- and structure-based virtual screening (VS)
  • An introduction to PyTorch for constructing, training, and evaluating neural networks
  • Theory underlying neural networks and common ways to improve their performance

This chapter marks a segway across several domains. In the first half of the book, we covered a sizeable swath of machine learning methods and how to apply them, typically using Scikit-Learn, to problems in drug discovery with a focus on ligand-based applications. In the second half, we will transition to advanced methods that underly deep learning. To apply these methods, we’ll expand our tool usage to PyTorch, a well-known and popular deep learning framework. Lastly, we’ll increase the scope of our projects to cover structure-based applications, e.g., how a ligand interacts with a known protein structure.

Figure 8.1 illustrates our roadmap for this chapter.

Figure 8.1 Roadmap for this chapter, starting with the high-level implementation of a PyTorch model and followed by detailed coverage of the underlying mechanics that govern our implementation.

8.1 Ligand-based VS with PyTorch

8.1.1 Protein Kinases

8.1.2 Our First PyTorch Model

8.1.3 Enrichment Factors in Virtual Screening

8.2 Neural Networks & PyTorch Mechanics

8.2.1 Models as Computational Graphs

8.2.2 Implementing a Neural Network

8.2.3 Training & Applying our Neural Network

8.3 Summary

8.4 References