Scientists at Google DeepMind have created an artificial intelligence (AI) program called AlphaMissense that can predict whether genetic mutations are harmless or likely to cause disease. The aim of this program is to accelerate research and improve the diagnosis of rare disorders. Missense mutations, where a single letter in the DNA code is misspelled, can disrupt protein function and lead to diseases such as cystic fibrosis, sickle-cell anemia, cancer, and brain development disorders.
Using AlphaMissense, the researchers evaluated 71 million single-letter mutations that could affect human proteins. With a precision level set at 90%, the program predicted that 57% of missense mutations were likely harmless, 32% were probably harmful, and the rest were uncertain in their impact. Based on these findings, the scientists have made the predictions available through a free online catalog to assist geneticists and clinicians studying mutations and diagnosing patients with rare disorders.
Currently, doctors have computer programs that attempt to predict disease-causing mutations but with limited accuracy, providing only supporting evidence for diagnoses. AlphaMissense outperforms existing “variant effect predictor” programs and can help experts identify disease-driving mutations more efficiently. Additionally, the AI program has the potential to identify previously unknown mutations associated with specific disorders and guide doctors towards improved treatment options.
AlphaMissense is an adaptation of DeepMind’s AlphaFold program, which predicts the 3D structure of human proteins. The AI program was trained on DNA data from humans and closely related primates to recognize whether missense mutations are common and likely benign or rare and potentially harmful. It also familiarized itself with protein language through studying millions of protein sequences and learning what a healthy protein looks like.
While the predictions made by AlphaMissense need verification from other scientists, the program shows promise in distinguishing between disease-causing and non-disease-causing DNA changes. However, some concerns have been raised regarding the complexity of the model and the need for a clear understanding of the AI’s workings in order to confidently make decisions in a clinical setting.
In conclusion, AlphaMissense represents a significant development in using AI to predict the impact of genetic mutations. If validated, this program may play a crucial role in disease diagnosis, allowing for faster and more accurate identification of disease-causing mutations.
Sources:
– Original Article: [Source]
– Science: [Source]
– Prof Ben Lehner: [Source]
– Prof Joe Marsh: [Source]
