All About Bioinformatics

Chapter 9: A machine learning approach to bioinformatics

Abstract

Bioinformatics algorithms had to be developed manually long before machine learning, which was difficult for issues like predicting protein structure. Instead of requiring the programmer to describe each characteristic individually, machine learning techniques such as deep learning may learn the characteristics of data sets. The algorithm can get better at combining simple traits into more complicated ones, and so on. Because of their multi-layered architecture, these systems can produce complex predictions when properly trained. Unlike previous approaches to computational biology, these methods prevent data from being interpreted and evaluated in ways that aren't expected. ...