New AI software has been developed that can compute a protein structure in just 10 minutes, saving years of research effort. This new Artificial Intelligence software is able to identify the three-dimensional structure of a protein from its amino acid sequence. The original system on which the new system is based, called Rosetta@home, was developed by researchers at the University of North Carolina’s School of Medicine and published in Nature Biotechnology on Jan. 16th, 2018.
Rosetta@home was based on the idea behind DeepMind artificial intelligence software. DeepMind was acquired by Google in 2014.
Researchers at Seattle’s Institute for Protein Design at the University of Washington School of Medicine achieved similar performance to that demonstrated by DeepMind in this project. The research team’s results were published July 15, 2021, in the journal Science.
The UW team used the UNC system, Rosetta@home in a method they called RoseTTAFold. Rosetta@home can be freely downloaded from GitHub by anyone, unlike the DeepMind system, which is proprietary, and owned by Google. More than 140 independent research teams have downloaded Rosetta@home since July. Scientists around the world are already using the data driven system to build protein models, greatly speeding up their own research.
Proteins are made up of strings of amino acids that exist in intricate microscopic shapes. The protein shapes provide valuable insight into how the proteins inside living organisms function. Scientists believe continued study of protein shapes will lead to improved treatments for many diseases.
David Baker, professor of biochemistry at the UW School of Medicine, director of the Institute for Protein Design led a team of computational biologists to develop the RoseTTAFold software. This intelligence software is an important ai platform because it greatly advances human intelligence, saving years of laboratory research that would be needed to understand the structure of each target protein. The software system uses “deep learning” to quickly and accurately predict the structure of the proteins, based on limited information.
Baker says the ai powered RoseTTAFold system computes protein structures in around ten minutes on a fast lab computer.
RoseTTAFold computed hundreds of new protein structures, including many poorly understood proteins from the human genome. They also generated structures directly relevant to human health, which not only included problematic lipid metabolism and inflammation disorders but cancer cell growth as well.
RoseTTAFold is a three-track neural network that solves the complex problem of taking into consideration three things:
- patterns in protein sequences
- how a protein’s amino acids interact
- a protein’s possible three-dimensional structure
Using one-dimensional, two-dimensional, and three-dimensional information flows, RoseTTAFold can reason the relationship between a protein’s chemical components and its folded structure.
RoseTTAFold’s ability to quickly perform these complex calculations is the reason it only takes a fraction of the time previously required to build models of complex biological assemblies.
The AI software has already contributed greatly to the understanding of the complex protein structures and may soon help to understand and overcome autoimmune and inflammatory diseases.
Rosetta@home Based on DeepMind
Rosetta@home was based on the ideas behind DeepMind artificial intelligence software. DeepMind was founded in London, England in 2006 by Demis Hassabis, Mustafa Suleyman and Shane Legg to solve general-purpose problems (e.g., not just games) using AI instead of relying on human expertise or knowledge.
The goal of the DeepMind founders was to solve problems that are beyond human intuition.
In 2020, DeepMind demonstrated outstanding progress in accurate protein structure prediction at the 2020 Critical Assessment of Structure Prediction (CASP14) conference. Since then, scientists have looked forward to accessing the ai software platform.
DeepMind and predictive analytics
DeepMind has been able to use predictive analytics to power advancements in fields like healthcare, energy and finance. DeepMind’s focus on AI comes with the belief that it will be used for more than just the deep learning to create computer programs. DeepMind has been able to develop AI that can solve tasks in an environment it has never seen before by using reinforcement algorithms, a type of machine learning algorithm.
DeepMind, AI Software and Society
The Deep Mind artificial intelligence system is not yet mature enough for general use outside research environments but this is DeepMind’s goal. DeepMind is confident that AI will be a positive for society and the company has ethical principles in place to make sure this happens, one of which includes transparency (Deep Mind).
So far DeepMind artificial intelligence has been able to use predictive analytics to power advancements in fields like healthcare, energy and finance.
DeepMind Machine Learning
DeepMind is one of the leading companies in Artificial intelligence research. DeepMind artificial intelligence has been used to help Google Deepmind’s games such as Atari Breakout, Pong, Space Invaders and Montezuma’s Revenge.
Deepmind was introduced to the world when it became the first AI machine to defeat a human at Go, an ancient Chinese board game, back in 2016. The AI beat Lee Sedol 4-1 in a five-game match. Deepmind’s artificial intelligence algorithms have also been applied to real-world situations such as climate modelling, disaster response, medical diagnosis and complex situations.
