Raygun AI Protein Design: Revolutionising Protein Engineering

Subject: GS 3: Science & Technology

Context: Recently, Scientists at Duke University have developed an Artificial Intelligence system called Raygun, that can shrink, enlarge, and extensively rewrite proteins.

What is Raygun?

• Raygun is an Al-based protein design system that modifies the length and amino-acid sequence of proteins.
• It uses protein language models trained on millions of protein sequences to learn the relationship between amino-acid sequences, structure and function.
• It can generate proteins of a target length in a single step, rather than repeatedly refining them.

How Does It Work?

• Fixed-resolution representation: Raygun divides proteins of different lengths into the same number of blocks and summarises each block.

UPSC Course Fees Online

• Probability-based generation: Instead of treating a protein as one fixed sequence, it models a range of plausible variants.
• Two control parameters: Users can determine how much the new protein should differ from the original and its desired length.
• High speed: A designed sequence can be generated in about 0.3 seconds on a single GPU, around 100 times faster than diffusion-based approaches.

Major Findings of the Study

• Wide Size Range: The researchers tested Raygun across a 17-fold protein size range, from haemoglobin with 147 amino acids to mTOR with 2,549 amino acids.
• mTOR (mechanistic Target of Rapamycin) Modification: The study found that mTOR could be shortened by more than 500 amino acids while retaining a largely intact predicted structure.
  • mTOR is a protein kinase that regulates cell growth, metabolism, proliferation and protein synthesis. The study found that it could be shortened by more than 500 amino acids while retaining a largely intact predicted structure. 
• Fluorescent Proteins: Researchers generated 70,000 candidates from each of two templates, eGFP derived from jellyfish and mCherry derived from coral.
• Experimental Validation: Computational screening reduced the candidates to eight, which were subsequently produced in human cells. Six variants retained fluorescence.
• Miniaturisation: The smallest functional fluorescent variants contained 199 and 206 amino acids, making them shorter than 96% of fluorescent proteins listed in the database used in the study.
• Extensive Rewriting: Some functional variants contained more than 40 coordinated insertions, deletions, and substitutions while retaining their function.
• Importance of Coordinated Editing: Random or poorly coordinated modifications can rapidly destroy protein function, whereas Raygun’s coordinated changes enabled substantial rewriting while retaining activity.

Potential Application

• Gene Therapy: Smaller proteins and redesigned biological components could potentially overcome size constraints in viral delivery systems.
• Drug Development: Protein engineering could enable the development of more compact and functionally optimised therapeutic proteins.
• Synthetic Biology: AI-based protein redesign can accelerate the development of new biological components and molecular tools.
• Basic Research: Compact fluorescent and reporter proteins could reduce the extent to which molecular tags interfere with the proteins being studied.
• Faster Protein Design: Raygun’s ability to generate candidates rapidly could reduce the time and computational resources required for protein engineering and screening.

Click to Explore UPSC Offline Coaching

Check Out UPSC CSE Books

Visit PW Store