‘Modelling’ of 6.17 million protein structures… Science, expectations grow
Metaplatform logo. random news
Meta Platform (hereinafter referred to as meta), the parent company of ‘Facebook’, has developed a computer program that predicts protein structure using artificial intelligence (AI), drawing attention.
The scientific community hopes that this program will facilitate the development of new drugs by helping scientists gain a deeper understanding of life processes.
According to a Wall Street Journal (WSJ) report on the 18th (local time), Meta’s research subsidiary ‘Meta AI’ recently developed ‘ESMFold’ and modeled 6.17 million protein structures.
“Predicting the structure of a protein can help scientists understand the biological function of that structure,” said study co-author Alexander Reeves, a scientist at Meta AI.
Previously, in November last year, Meta published a paper explaining the ESM fold on the ‘pre-publication thesis library’ (preprint server) platform.
The scientific community evaluated, “This program will help scientists deepen their understanding of life processes and advance the development of new drugs.”
In fact, Dr. Reeves that the EMS fold is already being used by academia, research institutions, and bio-companies.
The meta researchers produce protein structure predictions using a ‘Large Language Model’ (LLM) which predicts documents with only a few letters and words, leveraging the same technology Open AI’s ChatGPT produces answers similar to mankind. That’s what the WSJ explained.
Because proteins are constantly folding and refolding to form their final structure, determining their structure is difficult, expensive, and takes up to years using a microscope, but with AI, it can be done in hours or days.
Protein is a part of life and medicines are necessary for the functioning of cell tissues and organs Since medicines made from protein are used to treat heart disease, cancer and human immunodeficiency virus (HIV), it many pharmaceutical companies using AI to develop new drugs .are doing
ESM Fold Meta competes with AlphaFold, a protein prediction computer program from DeepMind, a subsidiary of Google’s parent company Alphabet.
Alphabet announced last year that it has 214 million protein prediction models in its Alphafold (DB) database that could facilitate drug discovery.
Meta explained that ESM fold is 60 times faster than alpha fold, but lags behind in accuracy ESM DB fold is much more than alpha fold because it creates a predictive model of genetic sequences that have not been studied from the front.
Andrew Ferguson, co-founder of Chicago-based biotech company Ebozyne and associate professor of molecular engineering at the University of Chicago, said Deep Mind and Meta’s AI predictive models are complementary as they have their respective strengths, and will lead to new discoveries in the future.
This research paper was published in the international scientific journal ‘Science’.
Lee Seung-goo Online News Reporter lee_owl@segye.com
[ⓒ 세계일보 & Segye.com, 무단전재 및 재배포 금지]
