And how we at Nexco can put it to work for you

Deepmind, the AI giant from Google’s holding, strikes back. Its new tool, AlphaMissense, harnesses the power of AlphaFold, the cutting-edge protein structure prediction AI, to predict the pathogenicity of missense mutations in proteins -presumably better than any other tool out there and with its predictions for the whole human proteome already available online. This new tool addresses a critical need in genomics and holds tremendous potential for the future of clinical diagnostics and precision medicine.

AlphaMissense tackles one of the most challenging issues in genetics: identifying the impact of missense mutations (i.e. mutations that result in amino acid changes in the coded proteins) on protein function. In principle, missense mutations can either disrupt protein function, which might in turn lead to diseases or not, or have minimal effect and be neutral. Deciphering the impact of mutations is vital especially for accurate diagnosing of genetic diseases. The breakthrough thus offers hope for unlocking the secrets hidden within our genetic code, ultimately benefiting life science researchers, health practitioners, and most important of all, patients.

AlphaMissense leverages AlphaFold’s ability to predict protein structures and combines it with a protein language model, enabling it to make precise predictions about the effects of missense mutations. The structural information from AlphaFold helps AlphaMissense identify regions where disease-causing mutations are likely to occur. By integrating the information coming from the protein language model AlphaMissense can then “decide” whether the amino acid change resulting from the mutation is likely to have a negative or neutral impact. The results speak volumes: AlphaMissense outperforms other computational tools in distinguishing disease-causing variants from benign ones and excels at identifying problematic variants identified in laboratory experiments.

Deepmind took its commitment a step further by creating a comprehensive catalog of all possible missense mutations in the human genome, available as raw data that we at Nexco can decode and interpret. Deepmind’s findings indicate that 57% of these mutations are likely benign, while 32% could potentially cause disease and 11% remain uncertain. This wealth of information opens new doors for researchers across multiple fields.

The potential applications of AlphaMissense are extensive. Molecular biologists now have a valuable resource for designing experiments that explore amino acid substitutions throughout the human proteome. Human geneticists could integrate AlphaMissense predictions into population cohort-based approaches to assess the functional significance of genes, particularly for shorter genes and low-frequency alleles where traditional methods may fall short. And last, most excitingly but requiring caution and a good amount of testing and benchmarking before actual applications, AlphaMissense could significantly improve diagnosis of rare genetic diseases, for which traditional methods just lack sampling.

While AlphaMissense represents a major leap forward, it’s important to acknowledge that this field is still evolving. As computational predictions become increasingly reliable, they will play a more significant role in diagnosing genetic diseases. However, rigorous evaluation, as advocated by experts, remains essential to ensure that these tools are ready especially for the sensitive application to real-world clinical applications.

AlphaMissense and Nexco

At Nexco we are up to date with how technologies continuously reshape the landscape of bioinformatics, chemical biology, and life sciences -and AlphaMissense is no exception. By directly harnessing the potential of AlphaMissense’s proteome-wide predictions, we can offer a unique advantage to our customers in leveraging this wealth of genetic information. For example, cross-referencing AlphaMissense’s predictions with our clients’ data to gain deeper insights into genetic variations, uncover new targets for drug discovery, or eventually (when the tool has been sufficiently benchmarked) to help optimize treatment strategies for precision medicine. Even simply browsing and querying the dataset, which is only available in a raw format at the moment, is something we can do for you.

Get to know more about us at https://www.nexco.ch/

References

Deepmind’s blog post on AlphaMissense: https://www.deepmind.com/blog/alphamissense-catalogue-of-genetic-mutations-to-help-pinpoint-the-cause-of-diseases

AlphaMissense paper in Science: https://www.science.org/doi/10.1126/science.adg7492

Comment on the paper in Nature: https://www.nature.com/articles/d41586-023-02943-5