Topic: Biology
Researchers at Stanford Medicine created two new AI models that can compare cells among species. The models were trained on data from 112 million cells representing 12 species, including humans and yeast. This breakthrough could lead to new insights into diseases and cell treatments.
Scientists at Stanford Medicine have developed two artificial intelligence (AI) models of the biological cell. The first model is called universal cell embedding, which paved the way for a second-generation model called TranscriptFormer.
TranscriptFormer has been trained on data from 112 million cells representing 12 species, ranging from single-celled yeast to humans. This allows researchers to easily compare cells among species and gain new insights into diseases and potential treatments.
The genome contains the instructions for life, but many biologists focus on gene expression – which genes a particular cell is using. For example, a beta cell in the pancreas expresses the gene for insulin, while white blood cells produce antibodies to fight disease. These differences can be used to tell cell types apart.
Researchers have built various atlases, or databases, that define cells and tissues based on their gene expression patterns. However, with tens of thousands of genes and hundreds of millions of cells from many species, it's difficult for humans to understand the data. The AI models help scientists process this complex information.
TranscriptFormer was trained in a similar way to large language models like ChatGPT. It looked at gene expression profiles and adjusted its parameters until it could correctly predict the missing values. This resulted in a 'universal space' – a mathematical space that can be used to measure how similar or different things are.
One use for TranscriptFormer is to explore evolutionary relationships between different species. For example, scientists who study sponges have wondered which kind of cell in a sponge is most closely related to a neuron in other animals. The model showed that the choanocyte cell in a sponge is similar to neurons in the roundworm and frog.
Why It Matters
This breakthrough could lead to new discoveries in medicine, such as understanding how diseases affect different species and developing new treatments. It also has implications for our understanding of evolution and the relationships between different species.
Key Facts
- Researchers at Stanford Medicine created two AI models that can compare cells among species.
- The models were trained on data from 112 million cells representing 12 species, including humans and yeast.
- TranscriptFormer is a second-generation model that uses gene expression profiles to understand cell relationships.
- The AI models help scientists process complex data and gain new insights into diseases and potential treatments.
- This breakthrough could lead to new discoveries in medicine and our understanding of evolution.
Key Terms
- Gene Expression
- The study of which genes a particular cell is using.
- AI Models
- Computer programs that can analyze data and make predictions.
- Universal Space
- A mathematical space that can be used to measure how similar or different things are.
Implications
This breakthrough could lead to new discoveries in medicine, such as understanding how diseases affect different species and developing new treatments. It also has implications for our understanding of evolution and the relationships between different species.
Leave a comment