Meta uses AI to predict protein structures

Meet ESMFold, the groundbreaking AI-based tool developed by Meta Platforms, Facebook's parent company. This amazing technology predicts the structure of a staggering 600 million proteins, paving the way for a deeper understanding of biology and possibly accelerating the discovery of new drugs. As proteins are the building blocks of life, they play a vital role in the function of tissues, organs, and cells, as well as being the basis for many medicines.

Imagine the potential! Using AI to predict protein structures could not only enhance the effectiveness of existing drugs and drug candidates, but also help uncover molecules that could treat diseases that have long been mysteries to scientists.

But ESMFold is not alone in this exciting race. DeepMind Technologies' AlphaFold also predicts protein structures with its own database of 214 million proteins. So, what sets ESMFold apart? It's an impressive 60 times faster than AlphaFold, although it sacrifices some accuracy for speed. Despite this, Meta's AI tool allows researchers to rapidly search through vast genetic databases, uncovering potential applications in medicine, health, food, and the environment.

The secret sauce behind ESMFold is the same large language model technology that powers OpenAI's ChatGPT. By using AI to predict protein structures based on known genetic sequences, scientists can gain valuable insights into the biological function of proteins.

While some experts prefer AlphaFold for its accuracy, others believe that both AI models possess unique strengths and will make significant contributions to new discoveries in the field. So, let's cheer on ESMFold and AlphaFold as they unlock the potential of proteins and help shape a healthier future for us all!