Sequencer giant Illumina launches AI algorithm based on primate genes to predict disease-causing gene mutations

As large models such as ChatGPT ignite a new wave of artificial intelligence craze, another "language" that has been accumulated for millions of years is also trying to use this technology to obtain the latest decoding. Recently, Illumina, the global leader in gene sequencers, announced the launch of a new artificial intelligence algorithm, PrimateAI-3D, in an effort to more accurately predict patient-causing gene mutations.

The total amount of genomic data generated each year is close to 40 billion gigabytes, according to data released by the National Institutes of Health. However, obtaining these data is only the first step in solving many mysteries of life. For them to be truly helpful to human health, in-depth interpretation and analysis are required.

Current scientific research shows that everyone carries millions of genetic variations. It is these variations that lead to individual differences in health and disease risk, but the way most variations work is currently unclear. The Human Genome Sequencing Center of Baylor College of Medicine and the Illumina Artificial Intelligence Laboratory hope to further solve the problem with the help of PrimateAI-3D.

According to Illumina, PrimateAI-3D uses a deep neural network architecture similar to ChatGPT and AlphaFold. The difference is that PrimateAI-3D is trained based on genome sequences rather than human language. "You can train generative language models like ChatGPT on existing text from Wikipedia and elsewhere. We use a similar deep learning architecture, but our data comes from millions of years of natural selection." Illumina Vice President of Artificial Intelligence President Kyle Farh said.

In addition, in generative language models such as ChatGPT, existing text can provide information for training, while the genetic variations that cause disease in the human genome are largely unknown. To solve this problem, PrimateAI-3D uses natural selection to train the parameters of a deep neural network. This training is based on the millions of benign genetic variants discovered by previously sequencing 233 different primate species. The largest sequencing effort of a non-human primate species to date.

Translation: We have shown that the more we know about genetic variation in non-human primates, the more accurately we can predict variants that may cause disease in humans. Rewrite: We have demonstrated that as our understanding of genetic variation in nonhuman primates increases, we are able to more accurately predict which mutations are likely to cause disease in humans. ” said Jeffrey Rogers from Baylor College of Medicine in the United States.

Subsequently, researchers from the Human Genome Sequencing Center and Department of Molecular and Human Genetics at Baylor College of Medicine in the United States, including Rogers, and the Illumina Artificial Intelligence Laboratory team led by Farh, applied the PrimateAI-3D algorithm to the British Biotech Identify potentially disease-causing human mutations among nearly 500,000 individuals in the sample library. Recently, two related studies were published in Science, a top academic journal.

They found that across 90 different clinical conditions investigated, 97% of healthy members of the general population harbored at least one highly modifiable variant lurking in their genomes. The findings also identified rare genetic variants that lead to higher risk for common diseases. Overall, PrimateAI-3D is at least 12 percent more accurate than any previous method in estimating genetic risks for health problems such as cardiovascular disease and type 2 diabetes, Farh said.

Farh added that one of the advantages of this new technology is that it can be used by all humans. It also means it overcomes the bias against people of white European ancestry inherent in existing genetic risk assessments, which are largely based on data from these groups.

Alex Aravanis, chief technology officer of Illumina, said, “The launch of this technology applies the latest artificial intelligence technology to genomics to reveal critical underlying information about complex genetic diseases such as diabetes, heart disease and autoimmune diseases. This brings huge opportunities to Illumina in genetic risk prediction and drug target discovery."

According to Illumina, PrimateAI-3D will be integrated into Illumina’s Internet software for use by the genomics community.

The above is the detailed content of Sequencer giant Illumina launches AI algorithm based on primate genes to predict disease-causing gene mutations. For more information, please follow other related articles on the PHP Chinese website!