Google DeepMind Releases AlphaGenome 2, Available for Non-Commercial Research
BIBIBI
AT A GLANCE
According to PANews, Google DeepMind has released AlphaGenome 2, which predicts DNA function based on specified tissue or cellular conditions and is now available for non-commercial research.
Article
Google DeepMind Releases Gene Prediction Model AlphaGenome 2
PANews November 11 news, according to the official Google DeepMind blog, Google DeepMind has released the gene prediction model AlphaGenome 2. The model can predict the function of DNA under specified conditions based on a DNA sequence of approximately 1000000 bases and a designated tissue, cell type, or state. Readouts cover gene expression, histone modifications, chromatin accessibility, and newly added DNA methylation. Unlike the previous generation, which predicted multiple tissues in parallel, the new model predicts for a single condition at a time, can distinguish between similar cell types, and can generalize to new single-cell data without retraining. DeepMind states that it identified nearly twice as many variants affecting cellular gene expression as the previous state-of-the-art model. AlphaGenome 2 is currently available for non-commercial research through the AlphaGenome API, while commercial research is in private testing...
(Click the link below to read the full article)
🔗 https://www.panewslab.com/zh/articles/01a1cb39-8116-701a-82ba-c71c8f0fee77
Key points
01
PANews reported on November 11, citing the official Google DeepMind blog, the release of AlphaGenome 2.
02
The model predicts DNA function under specified conditions based on a DNA sequence of approximately 1000000 bases and a designated tissue, cell type, or state.
03
Prediction readouts cover gene expression, histone modifications, chromatin accessibility, and newly added DNA methylation.
04
Unlike the previous generation, which predicted multiple tissues in parallel, the new model predicts for a single condition at a time; reports state it can distinguish between similar cell types and generalize to new single-cell data without retraining.
05
DeepMind states that the model identified nearly twice as many variants affecting cellular gene expression as the previous state-of-the-art model.
06
AlphaGenome 2 is now available for non-commercial research through the AlphaGenome API, while commercial research is in private testing.
AI-assisted interpretation
The following is analysis, separate from reported facts. Verify important claims independently.
It can be understood as a predictive tool that analyzes DNA based on specific tissue or cellular conditions: input a DNA sequence and specify conditions, and the model outputs functional predictions for those conditions. The changes this time include focusing on a single condition at a time and adding DNA methylation readouts; these outputs are predictions and should not be directly treated as validated experimental conclusions.
Why it matters to readers
The reported improvements focus on distinguishing between similar cell types and identifying variants that affect gene expression. If the relevant performance is validated, it may aid research into DNA function under different cellular conditions.
The key is to distinguish between "model predictive capability" and "experimentally validated results," and to note that availability for non-commercial research does not mean commercial research is fully open.
Risks and unknowns
The material is a media paraphrase with truncated text, and does not provide the complete official blog content or independent validation results.
"Nearly twice as many" is DeepMind's own statement; the material does not specify the comparison model, evaluation data, or specific methodology.
The scope and limitations of generalization without retraining are not specified.
The specific scope, opening timeline, and pricing arrangements for commercial research private testing are not specified.
Related Developments
Loading event timeline…
Related concepts
Gene Expression
This term is not in the glossary yet. Browse related concepts in the glossary.
