ISMB/ECCB 2025: Liverpool, LLMs and Lessons in Open Science

ISMB/ECCB 2025: Liverpool, LLMs and Lessons in Open Science

The Open Bioinformatics Foundation (OBF) Event Fellowship program aims to promote diverse participation at events promoting open-source bioinformatics software development and open science practices in the biological research community. Muhamad Haries Ramdhani, a PhD Student at the University of Aberdeen, was awarded an OBF Event Fellowship to attend the ISMB/ECCB 2025.

The Journey to Liverpool

I had heard about ISMB before, but I wasn’t entirely clear on its joint conference format with ECCB. What I did know was that ISMB/ECCB is arguably the biggest conference for computational biology in the world. For months, my attendance felt uncertain, as I initially didn’t know if the 2025 edition would be held in the UK. When the location was confirmed as Liverpool, it was a huge relief. Being a UK-based student meant I was fortunate enough to avoid a lengthy and often stressful visa application process.

As a PhD student, attending my first major international conference was a significant goal. My research focuses on benchmarking and cell-to-cell communication and the ISMB/ECCB program felt perfectly aligned with my work. I was particularly drawn to the DREAM Challenges track, as it offered a chance to see a wide variety of approaches to benchmarking, which is central to my own projects. Furthermore, the conference provided an ideal venue to present my own poster and receive feedback from experts in the field. Beyond the formal sessions, I was very excited about CollaborationFest. The opportunity to contribute directly to open-source projects and network with the developers behind the tools we use every day was a major motivation for me to apply and attend.

Keynote speakers and DREAM challenges talks at the ISMB/ECCB

Upon arriving, I realised the main conference didn’t formally kick off until Monday. Sunday’s schedule was composed mostly of tutorials and the student council symposium, both of which required extra payment to attend. However, the scientific energy began to build that evening with an incredible opening keynote from Nobel Prize winner Dr. John Jumper. In his talk, “Predicting the universe of biomolecular interactions with artificial intelligence,” Dr. Jumper walked attendees through the evolution of the AlphaFold project. He shared insights into the technical and conceptual milestones behind each generation of AlphaFold, including innovations like rawMSA masked language models, Evoformer modules and the importance of self-distillation. He emphasised that no single breakthrough carried the project, instead, AlphaFold’s success came from a steady accumulation of improvements and a methodology rooted in biological intuition and empirical iteration.

From Monday onwards, the conference was in full swing and the buzzing atmosphere truly hit. I have never been in a building with so many people. With over 2300 attendees, it was incredible to be surrounded by peers who all share a deep interest in computational biology. The sheer scale of the event was an experience in itself, requiring constant movement between auditoriums to attend the specific talks that aligned with my interests. My primary focus areas were the tracks for MLCSB, the NIH Track on GenAI, Cyberinfrastructure, Digital Twins and Quantum Computing and the DREAM Challenges. As my own research is in benchmarking, I spent the most time at the DREAM Challenge sessions. It was really nice to see the different and creative approaches people were taking to benchmark complex biological problems. I also made a point to attend talks in the Sysmod track, mainly because I was in a Systems Biology lab during my undergrad and the field still piques my interest.

The keynotes each provided a masterclass in a different corner of our field:

  • Amos Bairoch, in his talk “Plus ça change, plus c’est la même chose…”, delivered a powerful history lesson on 45 years of biocuration. He charted the field’s course from manually typing protein sequences in 1980 to leading the first Swiss-Prot release in 1986. Bairoch’s central message was a crucial paradox: despite incredible technological progress, the foundational work of expert curation still faces the same core challenges of being underfunded, undervalued, and wrongly assumed to be replaceable.

  • James Zou, the 2025 Overton Prize winner, explored the transformative potential of AI agents in his talk, “Computational biology in the age of AI agents.” He presented a compelling vision of AI “scientists” like the Virtual Lab, which can autonomously design research and CellVoyager, which reanalyses complex genomic data. His closing point emphasized that the future is not about replacement but a powerful synergy between AI’s scale and human creativity.

  • Charlotte Deane offered a dose of critical realism in her keynote, “Building the future of AI-driven structure-based drug discovery,” examining both the hype and the hurdles. While showcasing powerful tools from her own lab, she cautioned that the field is hampered by biased training data and flawed benchmarks. Her talk served as a crucial reminder that new AI models do not always outperform traditional methods on realistic test sets, raising serious questions about their true ability to generalize.

Posters, Serendipity and Collaboration

From left to right: Haries and his poster, Collaboration Fest and the code Haries was working on for Schema.science during the Collaboration Fest

The poster sessions were immense, with what I estimated to be around 500 posters presented each day in a massive venue. I came prepared with a list of posters I wanted to see and enjoyed many engaging conversations with the authors. The hall was also a hub of serendipitous networking. While wandering the aisles, I met a new friend, a fellow PhD student from Indonesia who is now at Nottingham.

Haries’ Collaboration Fest project working on the automation of generating training guides using LLM

The final day brought an event I had been looking forward to with great anticipation: Collaboration Fest. I have always wanted to join a Collaboration Fest and I was very excited for the opportunity to engage directly with the community in a hands-on setting dedicated to advancing open-source projects. It represented the perfect way to cap off a week of intense learning, shifting the focus from listening to actively participating and networking with developers to contribute to the tools that are so vital to our field. In Collaboration Fest, I worked with Phil Reed from schemas.science which is an initiative to improve the findability on the web of scientific research data, products and resources. I worked on improving the process of generating training guides for domain-agnostic schemas.science using LLM. I also had the chance to present my own poster on benCCChmarker, which is a software to benchmark multiple single-cell RNA-seq cell-to-cell communication algorithms. benCCChmarker provides an easy to use framework to compare different algorithms using simulated cell-to-cell communication single-cell RNA-sequencing data and curated data. I received a very interesting visit from people who work directly in the field, which led to a fantastic discussion about my work.

Lessons Learned and Reflections

Attending ISMB/ECCB 2025 was an invaluable experience that taught me several key lessons:

  1. The Scale of Community is Motivating: There is a unique energy that comes from being in a space with over 2300 people who share your specific scientific interests. The “buzz” in the venue was a powerful reminder that we are all part of a large, collaborative and global effort to advance science.
  2. AI is a Tool, Not a Panacea: The keynotes collectively painted a realistic picture of AI in biology. It is a transformative tool, but its success depends entirely on thoughtful model design, rigorous evaluation and most importantly, robust, well-curated data. Professor Deane’s talk was a particularly important reminder to remain critical and avoid being biased by hype.
  3. Foundational Work is Irreplaceable: Professor Bairoch’s talk was a crucial counterpoint to the focus on cutting-edge AI. It underscored that foundational resources built on expert human curation are the bedrock upon which new technologies stand. This work is vital, yet it continues to be undervalued and underfunded.
  4. Embrace Serendipity: While I had a schedule, some of the most memorable interactions were unplanned, like bumping into an old friend or starting a conversation at a poster. These moments are where new ideas and collaborations often begin.

This conference broadened my perspective on the challenges and opportunities in computational biology and gave me new ideas to apply to my own research. It was an intense, exhausting but ultimately rewarding week.

Acknowledgements

I am thankful to the OBF Event Fellowship for the travel award which allowed me to attend ISMB/ECCB 2025. My thanks also go to the University of Aberdeen for funding my PhD research and to my mentor, Dr. Michael Morgan, for his invaluable guidance and wisdom.