CRISPR-GPT for agentic automation of gene-editing experiments

Area of Science:

• Genomics
• Bioinformatics
• Artificial Intelligence in Biology

Background:

• Effective CRISPR gene editing requires deep knowledge of both the technology and biological systems.
• Large language models (LLMs) often lack the domain-specific knowledge needed for complex biological design problems.

Purpose of the Study:

• To present CRISPR-GPT, an LLM agent system designed to automate and enhance CRISPR-based gene-editing design and data analysis.
• To leverage LLM reasoning for complex task decomposition, decision-making, and human-AI collaboration in genome engineering.

Main Methods:

• CRISPR-GPT integrates domain expertise, retrieval techniques, external tools, and a specialized LLM fine-tuned on scientific discussions.
• The system assists users with CRISPR system selection, experiment planning, guide RNA design, delivery method choice, protocol drafting, assay design, and data analysis.

Main Results:

• CRISPR-GPT was used to successfully knock out four genes using CRISPR-Cas12a in a human lung adenocarcinoma cell line.
• The system facilitated the epigenetic activation of two genes using CRISPR-dCas9 in a human melanoma cell line.

Conclusions:

• CRISPR-GPT enables fully AI-guided gene-editing experiment design and analysis across different modalities.
• The system demonstrates effectiveness as an AI co-pilot for advancing genome engineering research.