Our AI writing assistant, WriteUp, can assist you in easily writing any text. Click here to experience its capabilities.

AlphaFold’s new rival? Meta AI predicts shape of 600 million proteins

View Original View Raw

Summary

The article discusses the recent development of a new AI-based tool for predicting the structures of proteins, which could potentially lead to new insights into biology. The tool, called ESMFold, is reportedly 60 times faster than existing tools at predicting structures for short sequences. The researchers who developed the tool say it could be used to help characterize proteins from unknown organisms, which make up a large portion of the proteins in the world.

Q&As

What is the "dark matter" of the protein universe?
The "dark matter" of the protein universe is the proteins from bacteria, viruses and other microorganisms that haven’t been characterized.

What are the benefits of using a language model to predict protein structures?
The benefits of using a language model to predict protein structures are that it is 60 times faster at predicting structures for short sequences and that it can predict the structures of proteins from metagenomic databases.

How many proteins does the ESM Metagenomic Atlas contain structural predictions for?
The ESM Metagenomic Atlas contains structural predictions for 617 million proteins.

What is the advantage of Meta AI's protein-structure-predicting AI over DeepMind's AlphaFold?
The advantage of Meta AI's protein-structure-predicting AI over DeepMind's AlphaFold is that it is about 60 times faster at predicting structures for short sequences.

What is the next step for such tools?
The next step for such tools is to find new kinds of RNA virus by looking for previously unknown forms of the viruses’ genome-copying enzymes.

AI Comments

👍 This is amazing news! DeepMind is really changing the game when it comes to protein structure prediction.

👎 I'm not sure this is a good idea. We don't know enough about these proteins and their potential effects on the body.

AI Discussion

Me: It's about a new AI that can predict the structures of 600 million proteins.

Friend: Wow! That's a lot of proteins.

Me: Yeah. It's a really big deal because there are a lot of proteins that we don't know anything about.

Friend: That's true. I wonder what implications this will have for science.

Action items

Use the ESM Metagenomic Atlas to predict the structures of proteins from bacteria, viruses, and other microorganisms.
Use the ESMFold model to predict the structures of proteins from metagenomic DNA.
Use the AlphaFold model to predict the structures of proteins from known organisms.

Technical terms

Proteins: Proteins are large, complex molecules that play many critical roles in the body. They are made up of amino acids, which are chained together in long chains.
DNA: DNA is a molecule that contains the genetic instructions used in the development and functioning of all known living organisms and many viruses.
RNA: RNA is a molecule that plays a central role in the regulation of gene expression.
Amino acids: Amino acids are the building blocks of proteins.
Metagenomic: Metagenomic refers to the study of genetic material recovered directly from environmental samples.
Computational biology: Computational biology is the field of science that uses computers to store, process and analyze biological data.
Bioinformatics: Bioinformatics is the field of science that uses computers to store, process and analyze biological data.