Google DeepMind has unveiled AlphaGenome, an innovative artificial intelligence model capable of reading and interpreting complex interconnections within human DNA. The technology allows scientists to peer into the least studied parts of the code of life, paving the way for understanding the nature of dementia, cancer, and rare genetic diseases. This was reported by the BBC, writes UNN.
Details
The human genome consists of 98% of the so-called "dark genome," which does not encode proteins but regulates the work of all genes. AlphaGenome is capable of analyzing a million letters of genetic code simultaneously, predicting the impact of mutations even in a single letter on the occurrence of diseases. This allows scientists to quickly identify the causes of obesity, diabetes, and hypertension, which are often hidden precisely in the poorly studied areas of DNA.
We view AlphaGenome as a tool for understanding what the functional elements of the genome do, and we hope that this will accelerate our fundamental understanding of the code of life.
Experts are already calling the model a "significant milestone" as it can accurately predict gene expression based solely on the DNA sequence.
AlphaGenome is already being used by over 3,000 scientists worldwide to find new drug targets. In oncology, AI helps distinguish random mutations from those that directly stimulate tumor growth. Although the model still needs improvement (particularly in predicting gene regulation over long distances), it has already become an indispensable assistant in synthetic biology and gene therapy development.
This success was a logical continuation of the work of the DeepMind team, which received the Nobel Prize in 2024 for the AlphaFold system. The new AlphaGenome model operates on the principle of "sequence to function," which allows not just reading the code, but understanding the biological consequences of each change in the human genetic text.