From code to control: scientists use AI to craft synthetic DNA

This image was generated with the assistance of AI
This image was generated with the assistance of AI

Leuven, 13 December - In a pioneering study published in Nature, Professor Stein Aerts and his team at VIB.AI and the VIB-KU Leuven Center for Brain & Disease Research describe a novel approach that uses artificial intelligence (AI) to guide the creation of synthetic DNA sequences. Their AI model was able to design functional enhancers, 'on switches' that can activate specific genes. The results represent a next step in decoding gene regulation and hold promise for future developments in gene therapy.

The cells in our body, from our eyes to our stomach, all have the same DNA. However, not all of this DNA is used in every cell type. The activity of genes is governed by a set of instructions within the DNA, called the regulatory code. Enhancers play a crucial role in this process by acting as control hubs dictating when and where certain genes are activated.

The logic behind the regulatory code has puzzled scientists for the past five decades. Now, a team of researchers led by Professor Aerts and first author Ibrahim I. Taskiran has been training a deep-learning model to crack it. This model was able to decipher the enhancer code, gaining unprecedented insight.

They then used this model to create synthetic enhancers tailored to specific cell types in the fruit fly brain. This approach not only worked remarkably well but also allowed the creation of alternative types of enhancers, including “dual code” enhancers that target two different cell types and extremely compact enhancers. Because of their small size and straightforward genetic manipulation, fruit flies are the first model organism of choice for many geneticists, however, the team was able to successfully extend their new approach to design human enhancers as well. 

Opening the black box

The study delves deep into the AI-guided design process of synthetic enhancers, offering a detailed understanding of their construction, nucleotide by nucleotide. This 'opening of the black box' provides unprecedented insight into the structure and composition of the enhancer code. The possibilities of these insights are numerous, ranging from basic biology to potential therapeutic interventions.

“The synergy between human insight and AI was key to navigating the enormous possibilities in enhancer codes,” says Dr. Ibrahim I. Taskiran. “Our deep learning model acted as a guiding oracle, allowing us to gain valuable insights into the enhancer code's structure and function.”
Prof. Stein Aerts (left) and Dr. Ibrahim I. Taskiran (right)
Prof. Stein Aerts (left) and Dr. Ibrahim I. Taskiran (right)

Tailored enhancers for gene therapy 

Understanding how enhancer activation is encoded in its DNA sequence is essential not only for modeling and predicting gene expression but also for the improvement of gene therapy.

That is why Professor Stein Aerts is enthusiastic about the implications of this proof of concept: “Scientists can use our approach to design cell-type specific enhancers and express any desired gene in any cell type they want. The possibilities of our insights are numerous. I have high hopes that our work will serve as a valuable and essential component in the field of gene therapy.”

Publication

Cell type directed design of synthetic enhancers. Taskiran et al. Nature, 2023. DOI: 10.1038/s41586-023-06936-2

The work received support from the European Research Council and Fonds Wetenschappelijk Onderzoek Vlaanderen. The study was published in Nature, back to back with complementary findings by a research team from the Institute of Molecular Pathology (IMP) in Vienna on ‘Targeted design of synthetic enhancers for selected tissues in the Drosophila embryo’. DOI: 10.1038/s41586-023-06905-9“. ​


India Jane Wise

India Jane Wise

Science Communications Expert, VIB

Joran Lauwers

Joran Lauwers

Science & Business Communications Expert, VIB


About VIB.AI

VIB.AI is the newly established VIB Center for AI & Computational Biology. The center's primary objective is to use machine learning to solve fundamental biology problems by combining it with in-depth knowledge of biological processes. Research areas range from developing foundation models and integrative theories of biological systems to creating innovative AI-driven biotech applications in synthetic biology, agro-tech, and personalized medicine. Several international experts are currently being recruited to expand the existing research expertise at VIB.

More information and vacancies: https://vib.be/ai

About the VIB-KU Leuven Center for Brain & Disease Research

Scientists at the VIB-KU Leuven Center for Brain & Disease study how brain cells are organized and how they communicate with each other. These mechanisms reveal and provide insights into what goes wrong in brain diseases such as Alzheimer's, Parkinson's, ALS, and dystonia. This basic work should ultimately lead to new drugs for use against these currently incurable diseases.

About VIB

VIB is an independent research institute that translates insights in biology into impactful innovations for society. Collaborating with the five Flemish universities, it conducts research in plant biology, cancer, neuroscience, microbiology, inflammatory diseases, artificial intelligence and more. VIB connects science with entrepreneurship and stimulates the growth of the Flemish biotech ecosystem. The institute contributes to solutions for societal challenges such as new methods for diagnostics and treatments, as well as innovations for agriculture. 

Learn more at www.vib.be.

Contact