news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

IT Home reported on August 22 that technology media The Decoder published a blog post yesterday (August 21), reporting that a study by Rutgers University showed that OpenAI’s GPT-4 language model can simulate simple amino acids and protein structures with high precision.

The relevant research results were published in Scientific Reports. The research team used the GPT-4 AI language model to explore its performance in basic structural biology tasks.The results showed that the AI ​​model can accurately predict molecular structures.

Scientists asked GPT-4 to build three-dimensional structural models of 20 standard amino acids. The feedback results accurately predicted the atomic composition, bond lengths and angles. However, GPT-4 made errors when simulating ring structures and stereochemical configurations.

In another experiment, GPT-4 was asked to simulate the structure of a common protein structural element, the α-helix, which required the integration of a Wolfram plug-in for mathematical calculations. The resulting model was comparable to the experimentally determined α-helix structure.

In addition, GPT-4 also analyzed the binding between the antiviral drug Nirmatrelvir and the main protease of SARS-CoV-2.The model correctly identified the amino acids involved in binding and accurately specified the distances between interacting atoms.

These capabilities stand out because GPT-4 was not developed specifically for structural biology tasks. The researchers note that GPT-4's modeling approach is not yet clear. It can use existing atomic coordinates from a training dataset or recalculate structures from scratch - further extensive research is needed to draw definitive conclusions.

Specialized AI tools such as AlphaFold 3 can predict more complex structures, the researchers said.GPT-4 is expected to complete basic structural biology tasks. This modeling capability is still rudimentary and has limited practical applications.

Nonetheless, the team says this study sets a precedent for applying this technology to structural biology. The researchers suggest that further research into the capabilities and limitations of generative AI could further explore the potential applications of AI in other areas of the life sciences beyond structural biology.