Vibe genes? Natural(ish) language comes to protein, genome design software
You can apparently now tell AI tools what sort of protein you want them to make.
TL;DR
- Sophisticated protein design tools previously required deep knowledge of protein structure and mechanisms.
- New AI tool Pinal translates natural language prompts into functional protein designs, simplifying the process.
- Pinal uses a two-step approach: T2struct generates a structural blueprint from text, and SAProt-T constructs the amino acid sequence.
- Training Pinal involved massive datasets of protein sequences, functions, and predicted structures from sources like AlphaFold.
- Testing Pinal showed a low, but not entirely discouraging, success rate in generating functional proteins like a fluorescent protein.
- Separately, Carbon X, an AI model from Hugging Face, enables DNA sequence design and analysis using natural language.
- Carbon X can run locally, offering cost and privacy advantages over commercial models.
- These tools democratize biological design but still require specialized knowledge and lab equipment for sequence validation and testing.
- The goal is to make complex computational biology tools more approachable for biologists, potentially accelerating research.