News

What's new at IQCB

Awards, new members, open positions and milestones from the IQCB community. Looking for dates? See Events.

14November, 2026

15 MINUTES AI FOR BIOSCIENCES: WATCH THE FIRST SESSION

Recording of 11 November 2026 · next session on Wednesday, 18 November, 9:30 CET

On 11 November 2026 we held the first 15 Minutes AI for Biosciences webinar. The topic was the /grill-me skill for Claude: instead of firing off one big prompt, you let the AI interview you first, one question at a time, until your plan, paper or project idea is clear. As the session put it: effort decides how far the AI goes, clarity decides where.

Recording, 13 minutes. It plays directly in the browser, also on phones and tablets.

Next session: Wednesday, 18 November 2026, 9:30 CET (8:30 UK time), then every Wednesday. 10–15 minutes, one topic, one question, back to work. Online and open to all at bbb.rlp.net (runs in the browser, no account needed). Anyone can present a result, a new tool or a practical AI tip: suggest and vote for the next topic at arsnova.eu. Add the series to your calendar.

11November, 2026

DESIGNING BIOLOGY WITH AI – SAFELY

Generative AI can now write DNA at genome scale. An interdisciplinary team from the Wübben Sandpit on AI in Biology, including IQCB junior group leader Dr. Maximilian Sprang, sets out in Nature Communications how safeguards can be built into every stage of the process.

Read more

Generative artificial intelligence can design biological sequences, from individual proteins to whole genomes. This opens new opportunities for medicine and biotechnology. But the same capability can be misused. How can the field capture the benefits while managing the risks?

That question was the focus of the Wübben Sandpit on AI in Biology in 2025, funded by the Wübben Stiftung Wissenschaft and organised by researchers of the Johannes Gutenberg University Mainz (IQCB, UMC) and the Robert Koch Institute in Berlin. It brought together experts in biology, artificial intelligence, biosecurity, ethics, medicine and policy from five countries and 20 institutions. In the foundation’s sandpit format, researchers from different disciplines spend several days working intensively on a shared question to develop new ideas and collaborations.

The participants have now published their results in Nature Communications. The paper sets out a defence-in-depth strategy that builds safeguards into every stage of the modelling pipeline:

  • the data used to train generative models,
  • the models themselves,
  • the synthesis of designs to physical DNA.

Why a single safeguard is no longer enough

Conventional LLMs are trained on human language. Genomic foundation models are trained directly on DNA, and have moved from prediction to design: published work has used such models to generate viable bacteriophages at the genome scale. The established line of defence – screening a sequence against databases of known hazards after it has been generated – becomes less reliable the more novel the generated sequence is. A sequence could be functionally hazardous and yet, because it resembles nothing on record, look unremarkable to similarity-based screening. Laboratory testing provides ground truth, but it cannot be repeated for every new model release.

Hence the layered approach. Rather than relying on one checkpoint, defence in depth distributes responsibility across the actors who can actually act on it: those who curate training data, those who build and release models, and the companies that synthesise DNA to order. Regulation is moving in the same direction. The EU AI Act, for instance, places obligations on providers of general-purpose models with systemic risk.

“The field of BioAI is exploding in the recent years and with increasing capabilities, researchers and people from policy began to ask for more safety measures. We see in the development of cybercapabilities by LLMs that the safety measures applied so far might not be enough and we’d rather not see this happen for biological AI models.”

– Maximilian Sprang

Mainz’s contribution

Few groups in Germany work at the interface of generative AI and biosecurity, while the topic is recently receiving considerable attention in the US and the UK. Dr. Maximilian Sprang, junior group leader at the Department of Dermatology at the University Medical Center Mainz and at the IQCB, is one of them. He was corresponding author of a related perspective on innovation and risk in the use of artificial intelligence in genomics, published in Trends in Genetics earlier this year.

His group is pursuing one of the strategy’s forward-looking layers: making the models themselves auditable. Instead of inspecting only the finished sequence, interpretability research could be used to inspect a model’s internal states while it generates — with the goal of safeguards that work in real time, during generation, or allow to keep human oversight in the loop. The IQCB, established at Johannes Gutenberg University Mainz as a centre for quantitative and computational biosciences, provides the computational infrastructure for this work.

The Wübben Stiftung Wissenschaft and the Robert Koch Institute are issuing their own announcements on the paper.

Publication:
Maximilian Sprang, Hassan Ahmed Hassan, Yasha Ektefaie, Anne Schmieder, Alix Wasilenko, Moritz S. Hanke, Rosa E. Martín-Peña, Rachael Harkness, Jasper Götting, Alan Murphy, Johannes U. Mayer, Marco Schäfer, André Friedrich, Kerstin Lenhof, Sophia Rudorf, Gigi Gronvall, Jan Kellmann, Simon Anders, Berkan Aysan, Nicole Saverschek, Kathrin Schuldt, Marc Güell, Katharina Ladewig, Georges Hattab (2026). A defence-in-depth strategy for securing AI driven biological sequence design. Nature Communications. DOI: https://doi.org/10.1038/s41467-026-78758-5 (available soon)

 

27April, 2026

IQCB MEMBER WINS COZZARELLI PRIZE

We are proud to announce that the prestigious Cozzarelli Prize in Biological Sciences was awarded to IQCB member Enrique Muro and his team during the annual NAS meeting on 27 April 2026. The study combines theoretical and observational approaches to understand how the genetic architecture of life was transformed during eukaryogenesis, enabling the emergence of higher biological complexity. Congratulations, Enrique!

Read more

27November, 2025

EINSTEIN FOUNDATION EARLY CAREER AWARD

We are happy to share that IQCB group leader Dr. Maximilian Sprang from the University Medical Center Mainz won the Einstein ECR Award in 2025!

Read more

26May, 2025

NEW MEMBER OF THE LEOPOLDINA

IQCB board member Prof. Dr. Hanna Kokko is a new member of the Leopoldina. Becoming a member of the German National Academy of Sciences Leopoldina is an outstanding achievement. Congratulations, Hanna!

Find out more about Leopoldina's mission here

JGU press release

 

23April, 2025

OPEN PHD POSITION

Computer-Aided Drug Design
Mainz, Germanydownload flyer

Really interesting PhD position in the field of drug design and machine learning available in Oscar Palomino-Hernández´s Lab at the University of Mainz in collaboration with the University Medical Center Mainz.

Application deadline: 9 May, 2025

Read more

 

18February, 2025

We are very pleased to announce that IQCB Members Miguel Andrade and Maximilian Sprang from the Faculty of Biology at JGU will be guest editors in a special issue of the MDPI Journal Genes.

They seek submissions focusing on computational aspects of batch analysis and/or correction in biological data. Their intention is to have a focus on multidimensional biological data, but they are open to expand to other types of data as long as they fall within the biomedical field.

Read more

Accepting submissions until 20 September 2025!

22August, 2024

AI PILOT FOR LIFE SCIENCES IN RLP

IQCB member Prof. Dr. Stefan Kramer was appointed AI pilot for the life sciences in Rhineland-Palatinate by minister president Alexander Schweitzer. Congratulations!

Read more

01August, 2024

*** NEW MILESTONE: 100 MEMBERS AT IQCB ***

25TH ANNUAL SYSTEM ADMINISTRATOR APPRECIATION DAY

26 July, 2024

We'd like to send a big, heartfelt thank you to our incredible system administrator, Robert Vettel!!!

05June, 2024

NEW STAFF MEMBER AT IQCB

Helge Vatheuer recently started his position as IT Liaison Manager

We're absolutely thrilled to introduce our newest team member, Helge Vatheuer! He is supporting us at the interface between research and IT infrastructure.

07May, 2024

In (theoretical) population genetic analyses for diploid organisms it is a customary assumption, which is also built into state-of-the-art statistical tools based on coalescent theory, that unlinked loci have independent evolutionary histories. This is conceptually doubtful because there is after all only one population pedigree, describing the set of reproductive relationships of all individuals for all time, and all genetic information is passed through that same pedigree. In particular, any gene genealogy must be a subset of this pedigree. It is known that for reasonably large and well-mixed populations this issue has little practical ramifications as long as the the variance of offspring numbers among individuals is not too large: the law of the gene genealogies conditioned on the fixed pedigree is then practically the same as its counterpart obtained by (a priori unjustified) averaging over independent copies of the pedigree.

In an article featured in Genetics 226(5), May 2024 https://doi.org/10.1093/genetics/iyae030  researchers including IQCB member Matthias Birkner https://www.staff.uni-mainz.de/birkner/ consider population models with potentially very high variability of offspring numbers among individuals. Here, the situation can be radically different, leading to "bursts" of coalescence triggered by the pedigree at certain times across all loci. This has interesting implications for ancestral inference and understanding of multi-locus data from population(s) (models) whose single-locus genealogies are multiple merger coalescents.

Full article

06February, 2024

NEW STAFF MEMBER AT IQCB

Robert-André Vettel now supports the IQCB as IT & Systems Administrator

Welcome Robert!!!

23January, 2024

IQCB OPENING EVENT

9:00 AM - ca. 3:30 PM Alte Mensa, Johannes Gutenberg University Mainz applications until 16.01.24

Please register here https://forms.office.com/e/DfFiB4zWRm. Registration will close on 16th January at 9 AM.

Link to Program

Menu