Introduction

Proteins are the fundamental building blocks of life, and the study of protein function is fundamental to fields such as biology, biochemistry, and medicine. To understand how a protein functions and how it interacts with other structures, it is vital to figure out its tertiary (three-dimensional) structure.

Traditionally, protein structures have been determined using onerous laboratory methods, including x-ray crystallography and nuclear magnetic resonance (NMR) spectroscopy. While these techniques are undeniably accurate, their cost and lack of scalability has led to significant bottlenecks in fields such as pharmaceutical development. Modern protein engineers need to determine large numbers of protein structures within days, not months or years. For this reason, computational methods of structure prediction are quickly becoming the preferred way to characterize protein structure.

Today, some software algorithms can predict the structure of a protein based on its amino acid sequence with accuracy rivaling these laboratory methods. With recent advances in computational biology, in silico prediction can be a fast, inexpensive, and reliable way to determine protein structures.

AlphaFold 2 was introduced in 2020 and was found in an objective worldwide test to be the most accurate algorithm for predicting the structures of single-chain proteins. AlphaFold-Multimer is a newer extension of AlphaFold 2 that was built to predict multiple chain protein complexes, also known as “multimers.” Both algorithms can be accessed online for free, but there are numerous hurdles for those seeking to leverage their benefits.

Chapter 1 describes the general steps involved in using software to determine tertiary protein structures. Chapter 2 discusses the advantages and challenges of using open-source AlphaFold 2 to predict single-chain protein structures or AlphaFold-Multimer to predict the structures of protein-protein complexes. Finally, Chapter 3 shows how NovaFold AI and NovaFold AI-Multimer provide an easier way to run and analyze predictions without needing a special computer or bioinformatics background.