by: Mark Cyril Mercado | Reprimo
Cracking the 50-year-old mystery of protein folding would solve many problems in molecular biology and open new scientific discoveries in biology. After all, proteins have various important functions correlated through their tertiary structure. For decades, determining protein structure relied on tedious and costly experimental methods, such as X-ray crystallography, nuclear magnetic resonance (NMR) imaging, and cryo-electron microscopy (Cryo-EM). As the adage says, determining the structure of one protein would take much work and time for one PhD candidate. Thus, the importance and gravity of solving the protein-folding problem.
One major breakthrough in the quest for understanding protein structure design was in 2003, when David Baker and his colleagues developed the Rosetta computer program. Baker and his team were able to engineer the 93-residue α/β-protein, named Top7, and validated its structure through crystallography. Baker attempted what was previously thought impossible—designing novel (de novo) enzymes from scratch with sequences unrelated to naturally existing in nature. This achievement opened up countless possibilities for future applications in biochemistry and biotechnology. Since its inception, Rosetta has continued to evolve, advancing in computational protein design.
Another enormous feat in protein architecture prediction emerged from the work of Demis Hassabis and John Jumper on an artificial intelligence (AI) system called AlphaFold2 (AF2). Originally developed by Hassabis’ DeepMind company, AlphaFold1 version reached an accuracy of 60%, beating all previous protein models when they first joined the 13th CASP (Critical Assessment of Structure Prediction) competition. At CASP 14, their team made another quantum leap when John Jumper joined and co-led the reformation to the new version: AlphaFold2. Under Hassabis and Jumper’s leadership, AF2 achieved an astounding 90% accuracy in predicting protein structures. After winning CASP14, AF2 was made to predict the protein structure of most of the human proteins and other sets of protein sequences from a variety of organisms. Notably, the source code of AF2 has been made publicly available, allowing researchers worldwide to explore, validate, and build on these models (https://github.com/google-deepmind/alphafold). A newer iteration of AlphaFold, AlphaFold3, also capable of de novo protein design and bimolecular interaction, is available on its server (https://alphafoldserver.com/).
In recognition of their revolutionary and brilliant contributions to computational protein design and protein structure prediction, the Royal Swedish Academy of Sciences has awarded the prestigious Nobel Prize in Chemistry 2024, with one-half of the prize to David Baker and the other half to Demis Hassabis and John Jumper. Their groundbreaking innovations have both advanced our understanding of protein folding and paved the way for applications in molecular biology.
SOURCES CITED
Åqvist, J. (2024). Computational Protein Design and Protein Structure Prediction. Nobel Prize. https://www.nobelprize.org/uploads/2024/10/advanced-chemistryprize2024.pdf
Fernholm, A. (2024). The Nobel Prize in Chemistry 2024: Popular Information. Nobel Prize. https://www.nobelprize.org/prizes/chemistry/2024/popular-information/
Posts
0 Comments