• Format: Face-to-face

• Dates: 4-5 September 2025 (two days)

• Location: Sozopol, Bulgaria, Hotel Villa List

​

This workshop will explore biological information through its dual nature: "digital" and "analog." These distinct modes of information processing in biological systems are deeply interconnected, collectively contributing to the regulation and functioning of living organisms. The "digital" aspect refers to genomic sequences —a linear, discrete coding system in DNA that operates via precise, rule-based codes (e.g., nucleotide sequences encoding proteins). This structure enables robust, error-corrected transmission of genetic information, allowing small, discrete units (genes) to be read, replicated, and passed on with high fidelity. Conversely, "analog" information arises from dynamic, continuous processes, particularly within signal-transduction pathways and cellular interactions. These pathways relay information in a graded manner, modulating responses to environmental stimuli, stress, or internal changes. The workshop seeks to advance understanding of biological information by emphasizing its dual nature and fostering synergy between theoretical and experimental perspectives.

Key questions include:

• How do theoretical models capture the interaction between digital genomic information and analog signal transduction pathways?

• Can a unified framework be developed to account for both types of information processing?

• How do findings in gene regulation, signaling networks, and epigenetics refine existing models of biological information?

• What roles do noise, stochasticity, and robustness play in these processes?

 

The workshop will highlight the potential of interdisciplinary collaborative research in addressing societal challenges, including AI development and applications, as well as public health, among other areas. These contributions aim to bridge the gap between research and real-world application, inspiring and motivating younger participants to pursue this complex and impactful career path.

​

​​

Keynote Speakers

​

​

​

​

​

​​​​

Ivan Coluzza (Rice University, Houston, US) 

​​​​

Biomimetic Polymers: Bridging Protein and Material Design

This presentation explores the exciting frontier of biomimetic polymers, which draw inspiration from the intricate world of protein structures to develop innovative materials like smart pores, nanoparticles, and plastics. By closely studying protein folding mechanisms—nature's own meticulous process for building complex, functional structures—we gain insights into creating artificial systems that mimic these biological properties. Our research delves into the simulation and design of biomimetic polymers that replicate key features of proteins, such as their unique ability to fold into highly specific three-dimensional shapes dictated by their amino acid sequences. This approach not only enhances our understanding of protein behavior but also paves the way for developing new materials with applications in medicine, technology, and environmental science. Through detailed studies, including numerical simulations of polymer translocation through nanopores and the development of surface coatings on nanoparticles, we illustrate how these materials can reduce unwanted interactions, such as protein aggregation, which is pivotal in medical applications. The promising results from both computational and experimental methodologies underscore the potential of biomimetic polymers to revolutionize the design and functionality of synthetic materials, drawing us closer to a future where materials are as dynamic and responsive as the biological systems they emulate.

​

​

​

​

​

​

​

​

Joanna Sułkowska (University of Warsaw, Poland ) 

​

Topology in Biological Matter: New Ways to Use Entanglement in Structural Biology

We have been aware of the existence of knotted and slipknotted proteins for over 30 years; however, the potential of entanglement in proteins has not been fully explored. From the perspective of chemistry and soft condensed matter, it is known that knots can perform various functions, such as increasing thermal or mechanical strength. During this lecture, I will show that the vast majority of knotted proteins have the simplest type of knot, and that the presence of knots is not preferred in any of the three domains—Bacteria, Eukaryota, or Archaea. Moreover, there is no organism without at least one knotted protein.

Recent advances in machine learning methods in structural biology have opened up new perspectives for protein analysis and prediction. I will demonstrate how machine learning techniques can be used to incorporate knotting into structural biology, enabling the design of fit-for-purpose new proteins that enhance or manipulate protein activity.  Additionally, I will discuss extensions toward the design of robust polymer–knotted-protein hybrid materials.

​

​

​

​​

​

​

​​

Noam Kaplan (Technion Haifa, Israel)

​

Learning Mechanistic Probabilistic Models of 3D Genome Organization

​

​Traditionally, the scientific method involves proposing quantitative mechanistic models and evaluating their predictive power. For example, we recently developed a probabilistic model, derived from a small set of mechanistic assumptions, which explains spatial chromatin interaction patterns that represent 3D genome compartmentalization. However, this scientific process is restricted and biased, as it is based on our partial knowledge and limited imaginations. How do we know if there is some better alternative mechanistic model? How can we know what specific assumptions need to be refined? On the other hand, black-box deep neural network models lack interpretability and do not typically provide mechanistic insight. Here we propose a principled approach to address these challenges by combining the interpretability of mechanistic probabilistic models with the flexibility of deep neural networks to study genomic data. As 3D genome organization has been proposed to be specified by a diverse set of stochastic mechanisms including liquid-liquid phase separation, ATP-driven molecular loop extrusion and polymer physics, we use a modular probabilistic modelling framework. This allows us to systematically replace individual model components with deep constrained neural networks, such that we can evaluate specific mechanistic hypotheses about distinct molecular mechanisms and how they relate to each other. Exploring this scheme, we remarkably find instances in which model components matching our hypothesis-derived mechanistic model are learned by the neural networks de novo, providing strong independent support. In other cases, refined or completely new explanations of the observed interaction data are provided by the neural networks. We also show how this framework allows discovery of mathematical equivalencies within model components, which can directly affect biological interpretation. Taken together, our work highlights how explicit probabilistic modelling can be tightly integrated with deep neural networks in order to test, extend and generate mechanistic models

​

​​

 

 

​

​

 

 

Stefano Piotto (University of Salerno, Italy) 

​

Beyond the Bottlenecks: Rethinking Drug Discovery from First Principles

​​

​The drug discovery process remains constrained by persistent bottlenecks, including reliance on heuristic-driven compound selection, protracted timelines, and escalating development costs. Current paradigms often depend on empirical screening and incremental optimization, limiting the pace of innovation. In contrast, emerging first-principles approaches integrating artificial intelligence and molecular simulation offer a transformative framework. These methods enable the rational design of bioactive compounds by predicting molecular behavior and therapeutic potential prior to synthesis, thereby reducing experimental burden and enhancing success rates. By coupling in silico design with data-driven modeling, this strategy facilitates the identification of novel chemical entities, including antimicrobial agents and biodegradable materials with pharmaceutical relevance. This rethinking of drug discovery from foundational principles holds the potential to accelerate development pipelines and address unmet clinical needs more effectively.

​

​

Expected Outputs

​

• Cross-disciplinary expertise transfer among participants

• Enhanced collaborative skills, including strategies for overcoming communication barriers between disciplines

• Expanded networking opportunities among diverse fields

• A digital repository for storing and sharing workshop materials

​

Travel and Accommodation

​

Sozopol is a town in southeastern Bulgaria, located on the Black Sea coast, 40 km south of Burgas. It can be reached by bus or taxi from either Sofia or Burgas airports.

• From Sofia: The distance is approximately 420 km. There is a direct bus from the Central Bus Station, and the ticket costs around EUR 25. Taxi fares vary depending on the car type and number of passengers, typically ranging from EUR 220 to 250.

• From Burgas: The distance is about 45 km. Taxi fares usually range from EUR 50 to 70.

Several hotels are located nearby, with Hotel Villa List, the Workshop venue, being the most convenient option. The hotel also offers airport transfers from Burgas Airport at a fixed price of EUR 50.

Please note that the Apollonia Art Festival takes place in Sozopol during the same week. Due to high demand for accommodation, we strongly recommend making a preliminary reservation as early as possible. If you are applying for financial support through the DYNALIFE COST Action, you may wait to confirm your booking after receiving the official invitation.

​

​

​

​

​

​