Unlocking DNA's Secrets with AlphaMissense

In some of our recent blogs, you may have seen us discuss how artificial intelligence has impacted the medical field, whether combating bacteria or even detecting genetic mutations in brain tumors. Well, AI has recently done it again in assisting scientists and doctors in detecting genetic mutations.

Imagine a world where AI can decode the secrets hidden within our genes, distinguishing between mutations that lead to disease and those that are harmless. Google DeepMind's latest creation, AlphaMissense, is making this vision a reality.

Recently, researchers introduced this tool, stating that it will significantly assist with researching rare and new diseases where it looks specifically for "missense" mutations. In some cases, a missense mutation can lead to a malfunctioning or non-functional protein, which may contribute to developing a disease or health condition. However, in other instances, missense mutations may have no noticeable effect or may even be beneficial, as they can result in proteins with new or improved functions.

Here are a few facts:

• A typical human has 9,000 mutations; they can be harmless or can cause diseases such as fibrosis or cancer.
• In recent studies, four million of mutations have been observed. However, only two percent have been classified as either disease-causing or benign.

(phys.org)

How does this tool work?

Data Collection:

AlphaMissense is trained on genetic information. This dataset includes DNA sequences from humans and information about how mutations in these sequences are observed to affect the corresponding proteins.

Pattern Recognition:

The tool is taught to recognize patterns in DNA sequences and their associated proteins. It then becomes proficient at identifying regular protein sequences and understanding a typical functional protein.

Mutation Analysis:

When AlphaMissense encounters a genetic mutation, it assesses the change in the DNA sequence and determines how it affects the amino acids that make up the protein.

Predictive Scoring:

AlphaMissense assigns a score to each mutation based on its analysis. The score indicates the predicted risk or potential harm associated with the mutation. As stated before, mutations are categorized as "probably benign," "probably pathogenic," or uncertain based on this scoring system.

While this tool is not intended for clinical diagnosis itself, it is a valuable tool for researchers and clinicians. It also can assist in increasing the rate of rare disease diagnoses. Furthermore, it can aid in discovering new disease-causing genes, indirectly contributing to developing new treatments.