By Richard Norman, Atte Sillanpää and Alexandre Bonvin
Our yearly Summer School on Biomolecular Simulations brings together young researchers from the computational biomolecular research community and internationally recognised leaders and code developers in the field. Now in its 7th year, the school was held over five days at the beautiful Sardegna Ricerche campus by the beach in Pula, Sardinia, which helps create a friendly and relaxed environment for learning and community building.
In this year’s post we are highlighting the community-building aspects which are so important for the success of the event year after year. Much effort goes into selecting the 30 participants from the high number of applications we receive. This is key to ensuring that we have a good blend of experience and diverse subject-matter expertise amongst the participants to create an atmosphere which favors networking and promotes exchange of ideas and discussions among like-minded people. Rather than taking our word for it, read some of the quotes from this year’s participants which confirm the value attributed to this aspect;
“I think the entire event was perfect. The number of people at the Summer School was ideal for getting to know each other while also having a variety of academic profiles. The time allocated for training and the time for breaks and recreational activities were very well organized. Honestly, I believe that everything together made it perfect. If I had to highlight something, it would be the great atmosphere among everyone and the eagerness and interest in sharing and learning from our projects. But I insist, the entire event has been wonderful.”
“As much as the lectures and topics were very interesting and relevant for my academic development, I think I enjoyed the networking the most, I met a lot of amazing people who made the Summer School an unforgettable experience.”
“I really appreciated the diversity of the topics covering most of the main aspects of biomolecular simulations, giving a comprehensive view of the Bioexcel softwares. Also organizers and speakers were really keen on discussing under the ‘no stupid questions exist’ framework and it was important to have the possibility of just getting to know and discussing science with really talented peers. And let’s not forget the beautiful location that really adds a lot to all the other aspects!”
“The whole spirit of the school was amazing.”
Based on this kind of feedback it is unsurprising that this year’s participants scored our Summer School with a 9.3/10 on average and actively recommended it for new researchers and students. We are pleased to report that 69% of the participants said they registered because colleagues had encouraged them while the remaining 31% had been reached via the BioExcel newsletter or social media. These data help to justify our ongoing efforts in the various BioExcel outreach channels.
To further promote engagement between all attendees, and for the first time this year, we did not limit the voting for the poster prize to in-house experts but rather extended it to all participants. The three poster prize winners are given the opportunity to present their work at a special edition BioExcel webinar and free registration to the upcoming BioExcel Conference on Advances in Biomolecular Research to be held in Brno, Czech Republic from 20-23 October.
Poster prize winners (L to R): Michael Mohm, Marta Devodier, Sachin Shivakumar.
Each year we follow and refine a tried and tested program containing a mix of lectures and hands-on tutorials covering the main topics in biomolecular modelling and simulation, in addition to having a high profile invited speaker to further enrich the program and inspire participants. This year we had the pleasure of hosting Ass. Prof. Sergey Ovchinnikov from MIT who gave participants some invaluable insights into AI-based structure prediction. Sergey explained the important concept of co-evolution at the basis of successful structure predictions, described the developments that have led to the success of Alphafold2 and related software, and gave a glimpse under the hood of their machineries. Sergey also talked about the very recent release of AlphaFold3, the new opportunities it provides but also its limitations. Sergey has been very influential in democratizing the use of AI-based structure prediction methods by providing the ColabFold resources.
Day 1 of the Summer School started with a lecture and hands-on tutorial combo on Molecular Dynamics by Alessandra Villa from KTH Royal Institute of Technology. The “Introduction to Molecular Dynamics” tutorial was run on interactive Jupyter Notebooks and you can access this through the GROMACS tutorials page.
Day 2 was all about integrative modelling and workflows, with Adam Hospital introducing the BioExcel Building Blocks (BioBB), which was followed up by a hands-on tutorial on how to perform protein-ligand complex Molecular Dynamics via Jupyter notebooks using BioBB. This tutorial, which uses GROMACS as the simulation engine, can be found together with other workflow examples on the BioBB workflow webpage. In addition to automating system preparation, simulation and analysis, BioBB enables your workflows, including the environment, to be encapsulated in a container making them reproducible and FAIR. This is particularly useful since scientific journals are starting to require reproducibility before publication and this is where BioBB can come to your aid. Our recent webinar on BioBB and FAIR workflows covers this topic.
During the afternoon session Alexandre Bonvin presented the various flavours of HADDOCK which can be used to predict the interactions between a plethora of distinct molecule types, including proteins, nucleic acids, glycans and small molecules and introduced the new modular version of HADDOCK (HADDOCK3), which allows for further flexibility and customisation of tasks. The long-awaited publication on the HADDOCK 2.4 webserver is now freely available. In the hands-on session, Victor Reys took participants through the HADDOCK3 tutorial for antibody-antigen modeling. This tutorial and other useful learnings including how to carry out antibody-antigen structure prediction from sequence using Alphafold can be found on the HADDOCK3 webpage.
Victor Reys introducing the antibody-antigen modelling tutorial.
The main topic for discussion during day 3 was Enhanced Sampling and its application. Erik Lindahl reminded participants to think and design their computational simulations carefully when attempting to test hypotheses and answer scientifically relevant questions. GROMACS’ implementation of various enhanced sampling methods can help support testing by: improving the efficiency of resource usage, sampling specific degrees of freedom and model dynamics on longer timescales, thus capturing slow transitions and conformational changes. The hands-on session on calculating the potential of mean force (PMF) along a reaction coordinate using the accelerated weight histogram method (AWH) in GROMACS helped illustrate one of the relevant methodologies. You can access this tutorial and view a recent webinar on its application.
Attilio Vittorio Vargiu, our BioExcel Ambassador for Italy, gave an in-depth brainstorming lecture on biomolecular recognition and enhanced sampling methods, covering techniques like Replica Exchange MD, Metadynamics, Accelerated MD, Targeted MD, and more, illustrated with multiple case studies. An engaging tutorial by Giuliano Malloci and Mohd Athar enabled participants to test the EDES method for enhanced sampling. The latest version of the EDES method (gEDS) which is designed to enhance the sampling of bound (holo)-like protein conformations, thereby improving docking performance and accurately predicting substrate-specific binding events for challenging biological targets was presented. Check out the latest publication on this topic.
During the last days of the school Bert de Groot provided an introduction to free energy calculations using the pmx software and this was followed by a hands-on tutorial by Sudarshan Behera on how to use pmx and GROMACS to set up alchemical free energy calculations for a ligand-binding relative free energy calculation study. You can access this tutorial and a webinar on the topic through our official YouTube channel. A recent benchmarking study highlights the possibilities offered by HPC systems in combination with GROMACS and pmx, demonstrating the computation of around 5,000 alchemical protein-ligand binding free energy calculations in one day. We finished off with an introduction to QM/MM by Emiliano Ippoliti, a ‘careers session’ for all participants and a final coffee in Pula’s beautiful settings.
This edition of the Summer School was a great success and we are already looking forward to next year’s event in June 2025 where we will welcome a new group of enthusiastic participants and aim to further strengthen the computational biomolecular modelling community in Europe.
