How the AI revolution is transforming the life sciences

2024 has truly been the year of AI. We have seen AI technologies embedded in everyday tasks like internet searches and recognised in the most prestigious awards for science and innovation. At EMBL-EBI, AI has been making a significant impact on our data resources, research, training, and more.

As the year comes to a close, let's take a look at some of the ways in which AI has transformed the life sciences.

AlphaFold recognised at the Nobel Prizes

This year, AlphaFold gained international recognition for its transformative impact on the life sciences. Developed by Google DeepMind, AlphaFold was trained on publicly-available data including the ones managed by EMBL-EBI.?

Together with Google DeepMind, we integrated AlphaFold 2 predictions into our data resources, making them easily available to all. AlphaFold has already been used to accelerate research on antibiotic resistance, vaccine development, plastic pollution and more.?

The AlphaFold Protein Structure Database that we co-developed with Google DeepMind provides predictions for almost all known proteins (over 200 million of them), and this abundance of data is freely available to everyone, everywhere.

Access the AlphaFold Protein Structure Database.?

“EMBL’s support in developing the AlphaFold Database was crucial; it significantly amplified the impact and reach of AlphaFold predictions across the global scientific community.”

– John Jumper, Director of Google DeepMind and Nobel Prize winner

The AI revolution hinges on accessible training

Using any technology to its full potential requires some experimentation and training, and AI is no different. This is increasingly true for AI tools that are being used by non-specialists, and a wide range of communities.?

EMBL-EBI Team Leader, Sameer Velankar, shares his thoughts on the topic including:

• Why it is important to provide accessible training for new AI technologies
• How to make training accessible
• Why AlphaFold users need training materials now

Make the most of AlphaFold2 with our free, online AlphaFold training course.

Empowering proteomics research with machine learning

Proteomics is another area where AI and machine learning applications have thrived this year. Introducing ProteomicsML, a collaborative, community-driven resource that streamlines proteomics data access for machine learning applications.

A free online platform, ProteomicsML is designed to simplify the process of preparing proteomics datasets for training machine learning algorithms. By acting as a centralised hub, it helps to make proteomics research more accessible and reproducible.?

Take a look at the ProteomicsML platform and find comprehensive tutorials for both experienced scientists and newcomers to the field.?

AlphaMissense data integration

EMBL-EBI has also worked on adding high-value AI predictions into our data ecosystem, when appropriate. For example, we integrated AlphaMissense data, developed by Google DeepMind, into several of our resources. These include Ensembl, UniProt, DECIPHER, ProtVar, and the AlphaFold Protein Structure Database.?

AlphaMissense classifies missense variants – point mutations that result in the substitution of one amino acid for another in a protein. By analysing protein sequences and structural contexts, AlphaMissense helps researchers to predict whether these variants are more likely to be pathogenic or benign. This is a useful tool for scientists trying to understand the links between genetic variation and disease.

“The deeper integration of AlphaMissense data into several EMBL-EBI resources is responding to requests from our user communities. This reflects how we work with our users more generally to add new features that empower scientists to gain new insights from data.”

We’re now also on Bluesky. Follow us to stay up to date with the latest EMBL-EBI news.