Scientists use AI to create music through proteins
First time a model of a human language was used for the synthesizing of proteins.
Lately, AI models have been flexing their muscle. ChatGPT, a platform that understands human languages, has recently become the poster child. A team of researchers have now tested a language to create amino acids sequences. This shows the ability to mimic human biology and evolve.
The language model named ProGen is capable of generating sequences of proteins with a degree of control. This result was obtained by teaching the model the protein composition. This is the first experiment to use a language-model to synthesize proteins.
The journal Nature Biotechnology published a study about the research on Thursday. The project was a joint effort between researchers from the University of California-San Francisco, the University of California-Berkeley, and Salesforce Research. Salesforce Research is the science arm of the software company in San Fransisco.
## The importance of using a model language
Researchers claim that a language-based model was used to create protein sequences with predictable functions across large protein families. This is similar to creating grammatically correct sentences in natural language on diverse topics.
Nikhil Naik is the Director of AI Research for Salesforce Research. He told *Motherboard* that amino acids are strung one by one to form proteins. The team used \”neural-language modeling on proteins to generate realistic yet novel protein sequences.\”
The study used 280,000,000 protein sequences, from more than 19,000 families. These were \”augmented with control tag specifying protein properties.\”
The team, according to Motherboard, uses conditional language models that allow them to have a greater control over the types of sequences generated. This makes it easier for them design proteins with specific properties.
Source:
https://interestingengineering.com/science/scientists-use-ai-to-create-music-through-proteins
Leave a Reply