Google's AI Tool Detects Genetic Mutations With 90% Accuracy

Human genomics is a complex field at the intersection of genetics, genomics, and medicine, dedicated to unraveling the intricacies of the human genome. It involves the study of vast and intricate genetic information within the DNA of individuals, exploring how variations and mutations contribute to health, disease, and individual traits. This field has revolutionized our understanding of genetics, enabling personalized medicine, disease prediction, and groundbreaking research into the genetic basis of complex disorders. Within the intricate realm of human genetics, approximately 9,000 genetic mutations course through the average human genome. While the majority of these mutations remain benign, a fraction can trigger conditions as severe as cancer, cystic fibrosis, or impair proper brain development (1^✔ ✔Trusted Source
Google AI tool predicts danger of genetic mutations

Go to source).

Advertisement

Uncharted Territory in Genetic Mutation Research

Remarkably, scientists have thus far identified a mere four million genetic mutations among humans, leaving an astonishing 71 million mutations largely unexplored. Of these, only a minuscule two percent have been categorized as either disease-causing or benign.

Amid this genetic complexity, Google's AlphaMissense AI tool emerges as a pioneering force. AlphaMissense specializes in making predictions regarding missense mutations—instances where a single genetic letter is inaccurately encoded within the DNA sequence.

Astonishingly, AlphaMissense exhibits a striking ability to predict these mutations, achieving an impressive accuracy rate of 90 percent while correctly identifying 89 percent of missense mutations.

Advertisement

AlphaMissense Analysis of Genetic Mutation

Researchers have harnessed AlphaMissense's predictive capabilities by establishing a precision threshold. When calibrated to 90 percent precision, the AI tool discerns that 57 percent of missense mutations are likely benign, while 32 percent carry potential harm, remaining uncertain about the remainder.

To equip AlphaMissense with its remarkable predictive powers, scientists fed it vast amounts of DNA data sourced from humans and closely related primates. This data allowed the AI to distinguish between common, benign missense mutations and rare, potentially perilous ones. AlphaMissense further honed its skills by analyzing numerous protein sequences, learning to recognize the characteristics of a "healthy" protein.

Upon analyzing a genetic mutation, AlphaMissense generates a score, effectively indicating the risk of that mutation causing disease, known as its pathogenic potential. In essence, this process parallels the understanding of language, where altering a word in a sentence can instantly signal a change in meaning.

Advertisement

Expansive Database as a Valuable Resource for the Scientific Community

The expansive database constructed by AlphaMissense has been made accessible to scientists, accompanied by a publication in the esteemed journal Science. This invaluable resource promises to facilitate further research and innovation in the realm of genetic mutations.

The scientific community views this breakthrough with optimism, anticipating that it may pave the way for new treatments and the ability to assess the risk of deadly mutations before they reach a critical stage. While not intended for standalone clinical diagnosis, these predictions hold potential for increasing the diagnosis rate of rare diseases and uncovering new disease-causing genes.

While AlphaMissense can predict the harmful nature of genetic mutations, it is important to underscore that the ultimate responsibility remains with human doctors. They must interpret the information provided by the AI tool and apply it judiciously in patient care and medical decision-making.

From Genes to Cures, the AI Tool is pledged to Advancing Healthcare

Reference:

1. Google AI tool predicts danger of genetic mutations - (https://phys.org/news/2023-09-google-ai-tool-danger-genetic.html)

Source-Medindia