type
status
date
slug
summary
tags
category
icon
password
DeepMind has released a new biological artificial intelligence system that specifically designs proteins that can bind to specific target molecules. These proteins can bind tightly to specific target molecules (such as viral proteins or cancer-related proteins). By generating these protein binders, AlphaProteo can accelerate progress in biology and health fields such as drug development, disease research and diagnosis.
AlphaProteo has demonstrated higher success rates and binding strengths than existing methods in experiments, especially for proteins related to the new coronavirus and cancer. The binding strength of protein binders designed by the system is on average 10 times stronger than the best existing methods. Some proteosomes are even better than proteins designed after multiple rounds of experimental optimization.
Key Features of AlphaProteo
Designing novel proteins
AlphaProteo can generate novel proteins that bind tightly to target molecules, which can serve as core tools in biological research and health research. For example, it can design protein binders associated with specific diseases, such as VEGF-A, which is of great significance for the study of cancer and diabetes. By generating highly efficient binders, AlphaProteo has shown great potential in disease research, drug development, and diagnosis. It can not only improve the efficiency of drug development, but also be used to design biosensors and accelerate basic biological research.
Multi-target design
AlphaProteo is able to design binders against multiple target proteins. It has successfully designed binders for target proteins related to cancer, inflammation, autoimmune diseases, and viruses such as SARS-CoV-2.
Versatile applications
The resulting protein conjugates can be applied in multiple fields, including drug development , disease diagnosis , cell and tissue imaging , and enhancing crop resistance .
Improved experimental success rate
AlphaProteo-designed protein binders show a higher success rate in experiments, reducing the need for repeated testing and optimization during the experiment, thereby speeding up the experimental process. For example, for some viral proteins (such as BHRF1), the success rate of laboratory testing reached 88%.
Targeting multiple protein targets
AlphaProteo has successfully designed binders targeting a variety of important target proteins, such as SARS-CoV-2 (new coronavirus) related proteins, BHRF1 viral proteins, and a variety of proteins related to cancer and inflammation (such as IL-7Rɑ, PD-L1, TrkA, IL-17A, etc.).
Features of AlphaProteo:
- High binding strength: The binding strength of proteins generated by AlphaProteo to target molecules is 3 to 300 times higher than that of existing methods, greatly enhancing the functional effectiveness of the protein. For example, the binding success rate for the target protein BHRF1 is 88%.
- Diverse binding: The system can handle a variety of different protein targets and use AI algorithms to design protein conjugates suitable for a variety of scenarios, including viral proteins and proteins related to cancer and inflammation.
- Experimental verification and optimization: AlphaProteo not only succeeded in computer simulations, but laboratory tests also confirmed the effectiveness of the proteins it produced in biological functions, such as preventing the SARS-CoV-2 virus from infecting cells.
- Continuous improvement and expansion: Although there are still challenges in working with some complex proteins (such as TNFɑ), the DeepMind team is working hard to improve AlphaProteo to tackle more complex design problems and further improve the algorithm's success rate and binding affinity.
Experimental Results of AlphaProteo
Successfully designed a variety of protein binders:
AlphaProteo has successfully designed binders for a variety of target proteins, including viral proteins and proteins related to cancer, inflammation, and autoimmune diseases. For example:
- SARS-CoV-2 spike protein receptor binding domain (SC2RBD): This is an important target protein of the new coronavirus. AlphaProteo has successfully designed the binders, and experiments have shown that these binders can effectively prevent the virus from infecting cells.
- BHRF1 viral protein: This is the target protein associated with viral infection. Experiments have shown that the conjugates generated by AlphaProteo have an 88% success rate, which is significantly better than existing methods.
Significantly improved binding strength compared to existing methods
AlphaProteo generated protein binders that were stronger 3 to 300 times than existing design methods on seven target proteins. For example, for TrkA (a protein associated with cancer), AlphaProteo designed binders that were stronger than binders generated by existing methods after multiple rounds of optimization.
High experimental success rate
In laboratory tests, AlphaProteo's combination showed a higher experimental success rate than traditional methods. This means that fewer designs need to be tested in the experiment, significantly improving experimental efficiency.
- BHRF1 viral protein: 88% of the designed binders were successful in the experiments.
- Other target proteins: such as IL-7Rɑ, PD-L1, TrkA, IL-17A, etc., showed high success rate and binding strength in the experiment.
Collaborative validation with the scientific community
AlphaProteo-generated conjugates have been independently validated by external scientists. For example, the team of Peter Cherepanov, Katie Bentley and David LV Bauer at the Francis Crick Institute confirmed that the SC2RBD and VEGF-A conjugates generated by AlphaProteo have the expected biological function. Specifically, some SC2RBD conjugates were able to prevent the new coronavirus and its variants from infecting cells.
Addressing challenges and limitations
Although AlphaProteo has been successful in multiple experiments, it still has limitations when it comes to more complex target proteins, such as TNFɑ, which is involved in autoimmune diseases. These proteins are more difficult to design binding partners for, but the team is continuing to improve AlphaProteo to address these challenges.
AlphaProteo has achieved remarkable results in the design of a variety of protein binders with high binding strength, and has demonstrated higher success rates and binding strengths than existing design methods in multiple experiments. These results provide new tools and technologies for drug development, virus prevention and cancer research.
Original article: https://deepmind.google/discover/blog/alphaproteo-generates-novel-proteins-for-biology-and-health-research/
- Author:KCGOD
- URL:https://kcgod.com/AlphaProteo
- Copyright:All articles in this blog, except for special statements, adopt BY-NC-SA agreement. Please indicate the source!
Relate Posts
Google Launches Gemini-Powered Vids App for AI Video Creation
FLUX 1.1 Pro Ultra: Revolutionary AI Image Generator with 4MP Resolution
X-Portrait 2: ByteDance's Revolutionary AI Animation Tool for Cross-Style Expression Transfer
8 Best AI Video Generators Your YouTube Channel Needs
Meta AI’s Orion AR Glasses: Smart AI-Driven Tech to Replace Smartphones