ERC Starting Grant for ten Utrecht-based researchers

Ten Utrecht-based reseachers have been awarded a Starting Grant by the European Research Council (ERC). ERC awarded the €1.5 million grant to eight researchers from Utrecht University (UU) and two researchers from UMC Utrecht (UMCU). The grant helps them to set up their own research project, assemble a research team and develop scientific ideas.

Projects and laureates

Roles of cytoplasmic bridges in cell-cell cohesion and communication

Cell bridges may have been the first step in the development of multicellular life.

The decisive moment in cell division, when cells actually split in two, is surprisingly variable in biology. At the end of cell division, the two cells are still connected by a bridge that they must cut, which some cells do quickly and others do not. Using her ERC Starting Grant, biologist Agathe Chaigne aims to uncover the roles of the long-lasting bridges.

Up until now, the bridge’s roles are surprisingly unclear. It could enable cells to exchange proteins, or to just keep cells together for as long as needed. To gain more insights into this, Chaigne’s team will study the bridge in stem cells and single-cell organisms called choanoflagellates, which can either live isolated from each other or as a group of cells.

“Cell division is one of the most interesting processes in biology,” says Chaigne. “Cells undergo really dramatic changes during division. Cells come in all shapes and sizes, yet all cells divide using the same basic processes and building blocks.” Uncovering the bridge’s functions could provide new clues for understanding how multicellular life evolved, and how organs in our body cooperate, says Chaigne. “The bridge may have been the first step in this development.”

Data diversity for fair and robust Natural Language Processing

If a language model learns certain stereotypes or performs poorly on specific tasks, it's often due to the data it was trained on.

Natural Language Processing (NLP) is a technology that enables computers to understand and use human language. Computational models in this field, such as the language model that ChatGPT uses, require thorough training before they can be effectively used. This training primarily involves large volumes of textual data. However, the focus now slowly shifts to the type of data that is used, says Dong Nguyen. “What topics are covered in the training data? Is it just news articles, or does it include other types of texts? Are different writing styles and dialects represented? Sometimes, a smaller, well-curated dataset can lead to better model training.”

In her DataDivers project, Nguyen aims to develop methods to accurately measure the diversity of datasets and investigate how this diversity influences the behavior of language models. Based on her findings, she seeks to create techniques for generating diverse datasets and training language models with diversity as a core component. Through this, she hopes to gain insights into how data diversity can contribute to fair and robust language models.

NLP is applied in various areas, such as creating chatbots, generating automatic translations, and detecting specific types of text, like hate speech. In recent years, the field has seen significant advancements, particularly with the rise of generative language models and their applications, such as ChatGPT.

Formalised Reasoning about Expectations: Composable, Automated, Speedy, Trustworthy

I think reinforcement learning, as a type of machine learning, will be applied more and more in the future.

Complex data analysis tasks require tools that automatically perform mathematical differentiation on the one hand, and programming languages that automate calculations with probabilities on the other hand. However, we lack a solid understanding of how to build systems that easily combine these two types of computations. This gap limits the capabilities for various machine learning and scientific computing applications, ranging from robotics to bioinformatics.

Assistant professor Matthijs Vákár aims to bridge this chasm by developing new theories and practical tools that allow for flexible and efficient calculations using both derivatives, which can help us solve optimisation problems, and probabilities, which help us quantify uncertainty. 

The new theories and techniques will, for instance, take a promising type of machine learning to the next level: reinforcement learning. Vákár claims this is important because many real-world problems are structured in a way that requires this technique. “I think this type of machine learning will be applied more and more in the future”, he says.

Vákár and his team will develop case studies in collaboration with domain experts, to ensure that they will focus on theory and systems relevant to real-world, complex modelling problems. Ultimately, the project will lay a trustworthy foundation upon which probabilistic data analysis applications can rise to the next level.

In-situ Mechano-catalysis for Polymer Activation and ConTrolled Conversion

Our experiments show that the idea works, but we don’t know exactly how. With this grant, we want to understand the fundamental chemistry underlying it.

Current plastic recycling generates plastic of inferior quality or requires a lot of energy and many refinery steps. Assistant professor Ina Vollmer developed a new recycling technique which uses force instead of heat to break down plastic. This way, the long polymer chains that make up plastics can be converted to monomers directly. Using force, instead of heat, is more sustainable and yields higher quality building blocks for new plastic products.

Vollmer and her team have already shown that the concept works in the lab. They did so with a groundbreaking experiment involving super strong marbles in a ball mill, together with pieces of plastic that needed to be recycled. When the mill rotates, the marbles grind the plastic until all that is left is a bit of white powder and a gas containing the monomers that can be reused to produce new plastics.

“However”, Vollmer says, “we don’t know exactly how this works. With this grant, we want to understand the fundamental chemistry underlying it.” This fundamental knowledge is needed to optimize and upscale these kinds of recycling methods. Vollmer will focus specifically on polyethylene and polypropylene, used to make, for example, shampoo bottles and yoghurt containers. 

PicAAA: Getting to grips with AAA+ ATPases encoded by positive-strand RNA viruses

This ERC project will allow us to journey inside infected cells and catch a glimpse at how viruses are built.

Viruses with small RNA genomes can have a significant impact on global public health. A prime example of such pathogens are Noroviruses, which in the Netherlands alone, sicken over half a million individuals each year. For vulnerable individuals, infections can be severe and potentially life-threatening. At present, no vaccines or antivirals are available to combat these infectious agents.

Challenging prospect

Despite their relatively simple makeup, small RNA viruses can completely rewire our cells during infection, transforming them into viral replication factories. Within these replication sites, viral proteins and hijacked cellular proteins form replication complexes that serve as platforms for the assembly of new virus particles. However, a detailed understanding of what these replication complexes look like is lacking, in part due to the difficulty of purifying these complexes. An alternative possibility is to study these complexes inside the infected cell, but the small size of these complexes makes this a challenging prospect.

Replication machinery

With the ERC project PicAAA, Daniel Hurdiss will study critical components of the viral replication machinery, with a particular focus on Noroviruses. “This ERC project will allow us to journey inside infected cells, and catch a glimpse at how viruses are built.” Together with his team, he will develop ways to assemble components of the replication machinery in a test tube, and also develop new methods to locate these virus assembly sites within the context of infected cells. Finally, molecules that interfere with components of the Norovirus replication machinery will be developed.

Pave the way for new therapies

This project will combine biochemistry, molecular virology, and state-of-the-art microscopy techniques to provide new fundamental insights into how small RNA viruses copy their genomes and package these into new virus particles. This project will pave the way for the development of new antiviral therapies.

RETIREWEL: Retirement across countries: A tripartite analysis of the welfare state, family care networks, and the retirement industry

This grant allows me to investigate two major and intersecting global trends: transnational mobilities and population ageing.

More and more people spend their retirement years across more than one country. This is for various reasons. They have a partner from a different country, they want to be close to family members abroad, they may have worked part of their lives in a foreign country, they are searching for more affordable healthcare, better climate amenities, and/or they wish to maximise their lifetime savings.

These complex mobilities create new challenges for national welfare regimes and families and produce new retirement markets. Still, we lack a clear understanding of the interaction between these different stakeholders in preparation for and during retirement. In the project RETIREWEL, Dora Sampaio addresses this puzzle through a novel, in-depth comparative study of transnational. She focuses on retirees and soon-to-be retirees, and examines these complex intersections in three migration corridors: Portugal-Brazil, Netherlands-Turkey, and UK-India. The project will function as a new knowledge base to understand the present and future of retirement and its implications for individuals and their families, institutions, and localities across countries.

Machine Learning in Science and Society: A Dangerous Toy? (TOY)

TOY increases our understanding of the promise and limitations of AI for producing (scientific) knowledge.

From medical science to fundamental physics, ever more scientific fields are turning to deep learning (DL) to solve long-standing problems or make new discoveries. At the same time, DL is used across society to inform and provide knowledge. 

Machine Learning in Science and Society: A Dangerous Toy? (TOY), a project that combines philosophy of science, epistemology, and ethics of technology, led by Emily Sullivan, will evaluate the potentials and dangers of adopting DL for epistemic purposes, across science and society.

TOY hypotheses that DL models are toy models. Toy models are highly idealised representations that simplify and distort complex real-world phenomena. Models like these are used in various scientific domains to ‘play around’ with, to gain insight into these complex phenomena. 

TOY is rethinking the nature and value of toy models and idealisation. In doing so, the project members aim to bridge gaps between the ethics of AI and the philosophy of science, and to provide insight into the appropriate use and trustworthiness of DL models in society. 

RESPIRE: Planetary Breathing in Asphyxiating Times

Your breathing is never simply ‘your own’ – the air we share are saturated with the suffocating consequences of planetary environmental challenges and social inequalities.

With every breath, we share the air with, for example, other humans, animals, plants, microbes, bodies of water, soil, and air. Planetary breathing and suffocation concern every process of living and dying on this planet. The project RESPIRE: Planetary Breathing in Asphyxiating Times, led by Magdalena Górska, will examine how the planetary breathing and suffocation are not purely environmental, but rather socio-environmental problems.

RESPIRE will focus on three ‘planetary lungs’ (forest, ocean, and soil) and specific problems they suffer (deforestation, dead zones, and peatlands, for example). The RESPIRE team will collaborate with natural scientists, organisations, artists, and stakeholders. Together, they will develop a socio-environmental approach, which will analyse how planetary breathing and suffocation reveal connections between social structures like power relations and environmental destruction.

RESPIRE is both an empirical and theoretical project. It will develop new ways of philosophical and artistic thinking, where planetary breathing and suffocation are not mere metaphors, but material articulations of current socio-environmental inequalities. As part of the project, Górska will set up RESPIRATORIUM. This research hub will bring together scholars, artists, and activists whose work addresses human and more-than-human breathing and suffocation in a norm-critical and intersectionally feminist manner.

Mapping brain fibers in living humans

Technical research can sometimes sound abstract, but it is crucial for making significant strides in the future.

Alberto De Luca aims to push the boundaries of imaging (MRI) to bridge a significant gap in neuroscience. Current MRI techniques can help us see how different parts of the brain are connected, but they are blind to what the wiring inside these brain regions themselves looks like.
Alberto will develop new MRI technologies to map the complex internal wiring of the cerebral cortex – where our cognitive processes take place – for the first time in living humans. This can help gain new insights into brain structures and better understand how aging or diseases like Alzheimer's affect the cortex, leading to cognitive impairments.

"I am incredibly pleased and honored with this ERC grant," says Alberto. “I have been dreaming about this project for two years, and this European support finally allows us to push the boundaries of what we can see with MRI, thus deepening our understanding of brain diseases and aging."

Human genetic determinants of deadly streptococcal infections

This ERC grant helps me link clinical observations in these patients to fundamental scientific discoveries about the cause of their disease.

Why do some otherwise healthy people develop life-threatening infections with the group A streptococcus bacterium, while most others experience only mild symptoms? This is the key question in the research of András Spaan. With his ERC grant, he will investigate the human genetic and immunological determinants in patients with these invasive infections.

The group A streptococcus is a bacterium that causes mild infections in most people. However, in some previously healthy individuals, it can lead to severe, life-threatening infections. Researchers and physicians do not understand why there is such a wide variation in infection severity among different people. András aims to investigate whether inborn, genetic defects of the immune system in severely ill patients explain their susceptibility to infections.

"Some people suffer from invasive group A streptococcal infections," says András. "It is a very serious infection; the bacterium is also known as the 'flesh-eating bacterium.' This project will hopefully contribute to the development of new therapeutic options to prevent and treat invasive group A streptococcal infections."