Deep Learning for Biology

Chapter 4. Understanding Drug–Drug Interactions Using Graphs

Graphs are a fundamental structure found everywhere in the world around us. A familiar example is social networks, where nodes represent individuals and edges capture the relationship between them. In train systems, nodes could represent stations and edges the routes linking them. Less obvious examples include research collaborations linked by coauthorship, web pages interconnected by hyperlinks, and supermarket baskets, where frequently copurchased items are connected.

Biology, too, is filled with data that naturally lends itself to a network framework—genes interact to control cell functions, proteins physically bind to each other, and cells send signals to each other, all forming graph-like systems. Even molecules can be represented as graphs, with atoms as nodes and chemical bonds as edges, as shown in Figure 4-1. At larger biological scales, ecological food webs capture predator–prey and other species interactions, while disease transmission networks map the spread of pathogens through populations.

Figure 4-1. Examples of graphs from different contexts. The social network shows people as nodes connected by edges representing relationships. The rail network illustrates stations as nodes and train routes as edges. The molecule network depicts the molecular structure of caffeine, where nodes represent atoms, and edges ...