Artificial intelligence (AI) is revolutionising how we approach and understand complex biological systems and changing the scientific landscape. From predicting protein structures to decoding genetic sequences, AI is not just a supplementary tool but a fundamental pillar of modern research.
As the world grapples with the myriad of possibilities AI brings, EMBL stands as a leader, exemplifying how to harmoniously integrate AI in research and through its delivery of data services. AI, intertwined with biology, has been both an area of innovation and an essential research tool for EMBL.
A legacy of data sharing and AI innovations
One of the pillars of EMBL’s AI capability is its open provision of life science data. These datasets are expertly curated and made freely available in the databases hosted by EMBL’s European Bioinformatics Institute (EMBL-EBI). This commitment to data sharing dates back to the foundation of the EMBL Data Library in the 1980s, which evolved into the European Nucleotide Archive (ENA). In the early 1990s, EMBL-EBI was established near Cambridge, UK, to help support EMBL’s dedication to data sharing.
The organised, curated datasets available through EMBL-EBI have been instrumental in AI developments by acting as training data for AI algorithms, including DeepMind’s AlphaFold algorithm for protein structure prediction. This AI-driven method has significantly expanded our understanding of protein structures across all known life forms. EMBL-EBI has also played a pivotal role in making the results of this groundbreaking work accessible via the AlphaFold Protein Structure database (AlphaFoldDB).
Bridging AI methods with biology
EMBL’s research community also actively innovates in the area of AI methods and their application in biology. The vast possibilities in AI-driven research are explored across EMBL’s six campuses throughout Europe.
We have methods developers, discovery data scientists, and innovative experimentalists all under one roof. We have closed the loop between experiment, data manipulation, and AI methods. Some notable instances of this integration include classifying entire organisms based on cell morphometrics, leveraging neural network embeddings to decipher developmental timelines, and exploring the underlying patterns in AI models — often termed ‘latent spaces’ — to better understand the influences of genetics. It is a rich world of possibilities, and I get a shiver up my spine each time I think about the future of this field.
Through AI, scientists can now tackle complex biological questions by converging diverse datasets, such as genetics and imaging. With the increasing fusion of AI into experimental design, the line between analysis and experimentation is blurring. Computational control in various molecular biology techniques can further be enhanced by AI-driven decisions. AI models, particularly those related to genome sequencing, effectively capture the nuances of biological data. To this end, groups at EMBL Heidelberg are pioneering advancements in ‘explainable AI,’ which promises clearer insights into complex biological systems.
Navigating the ethical landscape
Ethical considerations surrounding AI have become increasingly significant and cannot be sidestepped. In the realm of molecular biology research, the immediate repercussions and concerns stemming from AI might not be as immediately discernible as they are in other fields. However, as we see a convergence of research methodologies towards tangible clinical applications, the importance of identifying and addressing biases and ensuring equity rises to the forefront. Without action such biases could inadvertently influence outcomes, potentially leading to disparities in health care or treatment options.
Recognising the gravity of these challenges, EMBL’s Bioethics Services have taken a proactive stance. They are preparing for some of these discussions and we are engaging as topic experts in policy about these potential issues, but are also emphasising the need to frame these concerns within the broader societal context. This holistic approach ensures that technological advancements are balanced with their wider implications for individuals and communities.
The future of AI in biology
The future is undeniably promising for AI in molecular biology. EMBL is poised to remain a bedrock in this domain, championing responsible research and policy engagements, ensuring open data access, methodological advancements and discoveries using AI. As EMBL continues to train the next generation of scientists, it envisions a future teeming with AI-driven discoveries that will reshape our understanding of life.
The journey ahead is exhilarating, and EMBL is firmly positioned to lead the way in this AI-augmented exploration of life sciences.