One of the "grand challenges" in the field of Biology has found resolution after 50 years of existence as the artificial intelligence (AI) software called "AlphaFold" developed by AI company DeepMind solved the challenge. Now, structures of proteins being folded could be accurately predicted by the AI.
As thousands of proteins surround and build up the living cells inside bodies, determining the shape of a protein undergoing a protein folding process is very vital and important for the world of science and technology today. It helps studies involving various diseases, such as cancer and dementia.
The AI system championed a competition called the Community Wide Experiment in its 14th installment. It is organized and facilitated by the Critical Assessment for Structure Prediction or CASP, which dedicated their focus on the long-existing matter of protein folding structures over the past 26 years and counting.
The Community Wide Experiment for this year was notched by DeepMind. As CASP described, AlphaFold reached and achieved "unparalleled levels" of protein structure prediction accuracy by presenting and delivering 3D protein structure predictions on the most well-calculated and accurate level.
CASP chairman John Moult announced the news of DeepMind's victory by a press call, stating that a 50-year old challenge for the field of computer science and biology "has been to a large degree solved." He also mentioned that there will be significant impacts for the field of drug design and manufacturing because of their expected benefit from the system of protein folding prediction now being solved by AlphaFold.
DeepMind's Journey on AlphaFold
DeepMind's research on protein folding started years ago. The company first joined the Community Wide Experiment in 2018, yet even with fascinating results and outcome, they did not make it to the final cut and has not seen top marks from the competition organizers themselves. But that did not stop the company from expanding further their research on the five-decade problem, which would then be eventually solved by them after two years.
Yet like any other company, they also found themselves on edge. The company's founder and Chief Executive Officer (CEO) Demis Hassabis stated the company's progress along the way on developing AlphaFold. He remembered the times where they have put up so much information, but also there are times that progress was not approaching them. According to him, there were three months that they went with zero progress. He also said that even when the competition's deadline fast approached, errors and mistakes were still happening during their trials and test runs.
But all of these seemed worth the trials and hardships after capturing the gold of the Community Wide Experiment. He sees their achievement not just their own achievement, but also for the scientists and for the continuing world of science and research, particularly the protein folding process. As he reiterates, "The reason you have a structure is to understand something about the natural world and then ask even more questions."