Revolutionizing Protein Design
Researchers at Shanghai Jiao Tong University have made significant advancements in protein design by leveraging artificial intelligence. They have established the world’s largest protein sequence dataset and designed innovative models based on this rich dataset. This pioneering work enables targeted modification and selection of proteins with specific functionalities.
Efficiency Redefined
According to the research team, this advancement can drastically reduce the time and cost associated with industrial protein modification. Traditional methods relied heavily on lengthy trial-and-error processes, which typically consumed two to five years. Now, the cycle has been reduced to just six to twelve months, enhancing efficiency tremendously.
Real-World Applications
Proteins play crucial roles in various industries, from pharmaceuticals to environmental sciences. For example, in laundry detergents, proteins must withstand temperature variations to effectively decompose stains. The researchers’ innovative approach transforms protein production, allowing for directed modifications that make proteins more resilient and suitable for specific applications.
The Venus-Pod Dataset
The team has also developed the Venus-Protein Outsize Database, or Venus-Pod, which contains over 9 billion protein sequences from diverse organisms, covering terrestrial and marine life forms and extremophiles. Notably, 500 million of these sequences are tagged with functional attributes detailing their operational characteristics.
Leading the Industry with AI
The Venus series models, built on the Venus-Pod dataset, focus on accurately predicting and designing protein functions. Lead scientist Hong Liang explained, ‘Our models optimize underperforming proteins and identify those with exceptional functionalities. These unconventional proteins have the potential to revolutionize biotechnology and pharmaceutical industries.’
Automation Enhancing Research
In addition to these models, the team has created an integrated machine capable of performing over 100 protein expression and testing tasks within a 24-hour period. This advancement boosts efficiency nearly tenfold compared to manual methods, reducing labor and resource costs significantly.
Successful Applications
Within two years, the Venus models have been utilized to design proteins for practical applications, particularly in the early diagnosis of Alzheimer’s disease. The modified alkaline phosphatase developed by the team shows three times the activity of existing products, allowing for the detection of disease biomarkers at extremely low levels.
Looking Forward
This achievement marks a critical step in the evolution of biotechnology, promising to pave the way for future innovations in protein engineering and synthetic biology.
