Updated: Mar 15, 2021
Protein Folding: What is it?
Proteins are multi-functional molecules that form the building blocks of life. Proteins are synthesized by the cellular systems, from amino acids. Chains of amino acids are folded into a particular structure before they can be assigned a specific function. This phenomenon is called protein folding, and determines the unique pattern of a protein. In case of misfolding, accumulation of a malformed protein can result in various pathological conditions.
Conventional Methods in Proteomics
For more than half a century, X-ray crystallography has been the main modality utilized for estimating correct protein structures. However, this conventional method is quite expensive, and it can require as long as a year to determine the exact structure of a single protein. Another technique that has been adopted is transmission electron cryo-microscopy. However, it won’t be long before these two outdated methods become replaced by the modern scientific innovations discussed below.
Artificial Intelligence in Protein Structural Analysis
Recently, a Google-run company, DeepMind, launched an Artificial Intelligence (AI)-based software, that allows for more accurate prediction of a protein structure. The scientific program, called “AlphaFold”, is based on Machine Learning, which is a major component of AI. The system works by analyzing the various patterns of amino acid residues, and other parameters that constitute the input information. Through a highly intricate, built-in algorithm, Machine Learning can estimate the correct distance between any two constituent amino acids of a protein. In this manner, the entire protein structure can be predicted with a high fidelity.
How Accurate is AlphaFold?
The most recent version of AlphaFold has proven its remarkable efficiency at the Critical Assessment of Structure Prediction (CASP) challenge. Its score was recorded to be around 90%, which gave it a significant lead over other competitors of the CASP challenge. This astonishing feat proves that AlphaFold’s AI system can be even more reliable that crystallography, cryo-microscopy, and other methods used for the structural analysis of proteins.
How can it be Helpful for Biological Research?
AlphaFold can substantially transform the field of molecular research in biology. In the near future, it will be possible to achieve a drastically higher efficiency and precision when determining protein structures. Most importantly, we can minimise the time taken to determine these structures. Experimental data have shown that the protein structures that usually take years to crack, have been accurately determined by AlphaFold in less than an hour.
Following are a few potential advantages that can be derived from this AI program:
A better understanding of the pathogenesis of various diseases that arise due to protein misfolding.
Early and accurate estimation of bacterial and viral proteins which can help modulate the course of antimicrobial therapy.
Applications in tissue regeneration.
Misfolding proteins produce prions, which are extremely dangerous substances, that cause a variety of diseases. Here's a short video that explains this more:
While AI continues to transform the world of proteomics, the success of AlphaFold is bound to have a lasting impact on other aspects of biology, and neurology. Soon enough, we will witness the marvels of AI and Machine Learning on a much broader scale.