A new artificial intelligence (AI) called “ESMFold” has been developed by Meta (Facebook) that can accelerate the prediction of high-resolution protein structures by up to 60 times faster than the previously leading AI “AlphaFold 2.0” from Google Holding Alphabet. Despite providing relatively inaccurate results, the deviations are minor.
Knowing the 3D structure of proteins is critical information for biology and pharmacy. Proteins are essential bio-machines in our bodies that perform various functions such as building material in hair and nails, hormones, and antibodies. Prediction of protein structures helps to determine their biological function in the body, potency as drugs, and suitability as drug targets.
ESMFold determines the 3D structure directly from the sequence of the amino acid building blocks, which is encoded in the base sequence of the DNA. The AI finds patterns in the amino acid sequences that correlate with specific structures, filling gaps in amino acid chains and determining the most probable amino acid for missing positions. ESMFold no longer needs to compare the investigated amino acid sequence with other amino acid sequences with a known 3D structure, unlike AlphaFold and RoseTTAFold, developed by the University of Washington.
ESMFold researchers have predicted far more protein structures with high reliability thanks to the higher speed. According to the researchers, the method enables prediction of the structure of previously unknown proteins from environmental samples and can be applied directly to predict the consequences of point mutations.
In summary, ESMFold can usher in a new era of more advanced protein structure prediction and provides novel insights for biomedical applications.
