The 'Raygun' Breakthrough (Science and Technology)
The 'Raygun' Breakthrough (Science and Technology)
Researchers at Duke University School of Medicine have developed an AI system called 'Raygun' that can shorten, expand or extensively rewrite existing proteins while preserving their structure and function.
Source: The Indian Express, Page 9, 10 August 2026, 'Scientists can now shrink and rewrite proteins with AI'
What Does Raygun Do?
Unline Protein Language Models (pLMs) which capture the “grammar” of proteins, Raygun focuses on compressing and regenerating protein sequences at different lengths.
Fixed-Size Representation: Raygun divides every protein sequence into a fixed number of blocks and summarises each block, creating a representation of the same size for all proteins.
Information Compression: The model is trained to preserve sufficient biological information in these summaries so that the original or modified protein sequence can be reconstructed.
Probabilistic Generation: Instead of producing a single fixed representation, Raygun models a range of possible protein variants, allowing it to generate new sequences at a desired length.
Why Protein Design Matters
Proteins, built from chains of amino acids, drive nearly every biological process, from wound healing to fighting infection.
Traditional protein engineering relied on small, one-amino-acid-at-a-time edits, since possible combinations explode exponentially as changes increase.
Applications include gene therapy, drug development, medical imaging, industrial enzymes and biotechnology research.
Protein Language Model (pLM)
A Protein Language Model (pLM) is a type of artificial intelligence designed to analyse and understand proteins.
Similar to tools like ChatGPT or translation software, it works on protein sequences instead of human language.
It treats amino acids as “words” and the entire protein sequence as a “sentence”, allowing it to interpret biological information.
pLMs learn protein “grammar” by training on large databases of protein sequences and identifying patterns and relationships between amino acids.
pLMs can predict missing amino acids in a protein sequence, similar to how language models predict the next word in a sentence.