What changed
Hugging Face, the AI platform, released a new open-source model called Carbon-X on Thursday. This model is designed to understand DNA sequences and answer questions about the effects of DNA changes on gene activity and health. It can run locally on a laptop, without requiring cloud-based processing.
What this means in practice
The model was developed in collaboration with researchers from Zhongguancun Academy in China and the University of Naples Federico II in Italy. It is trained on genomic sequences and natural language, allowing users to ask conversational questions about DNA. Lewis Tunstall, a senior research scientist at Hugging Face, explained that the tool is intended to help researchers design new drugs and genetic therapies, and could also assist patients in interpreting their own genomes.
The model integrates DNA and natural language in a conversational interface, similar to ChatGPT for DNA. It was trained on 300 billion tokens of data, including DNA sequences from various organisms and English text.
However, users cannot upload full human genomes, as the model is trained to process DNA sequences of up to 50,000 base pairs. Tunstall also mentioned that the team did not train the model on viruses to avoid potential misuse.
Limitations and availability
Carbon-X is a small model with 14 billion parameters, designed to run locally on user hardware. This ensures that no data is sent to Hugging Face or any other company. The developers caution that they have not validated the model with actual wet lab experiments, so they cannot confirm the accuracy of its predictions. The model is available through Hugging Face's platform for developers, researchers, and others interested in experimenting with AI without building everything from scratch.
