Online Seminar Series
Modelling approaches for disease processes
The Online Seminar Series (Zoom) “Modelling approaches for disease processes” is organized by the e:Med project group Modeling of Disease Processes.
Time: Every 1st or 2nd Wednesday of the month, 2 p.m.
Aim/ Focus: Introduction, discussion and comparison of different mathematical modelling approaches. For initial list of topics and speakers see below.
Duration: 30’ talk +15’ discussion. We would like to have a focus on discussion. In cases of 2 speakers the duration can be 40 min + discussion.
Targeted audience: PhD and master students, postdocs, group leaders in the modelling field.
The talks start at 2 p.m. CET
If you missed a seminar, you can watch the video recordings of several the talks and discussions. Please see below in the 'previous seminars'.
* Talks with video are marked by an asteriks.
We look forward to the new insights into disease process modeling and to discussing them with you.
Nov 8, 2023 - Computer models of tissue regeneration through agent-based modeling approaches. Sara Checa
Prof. Dr. Sara Checa, Julius Wolff Institute, Berlin Institute of Health (BIH), Charité - Universitätsmedizin Berlin
Location: online ZOOM
Computer models of tissue regeneration through agent-based modeling approaches.
Several tissues within the body have the amazing ability to self-regenerate, e.g. skin, bone. Understanding the mechanisms behind this process might serve as a fingerprint to foster regeneration in other tissues that have none or limited regeneration capacity. The regeneration of injured tissues within the body is a highly complex process, where different cell phenotypes interact in a highly coordinated manner. In this process, cells migrate, proliferate, differentiate and deposite new extracellular matrix over time; ideally leading to the reconstitution of the original tissue function. In this talk, I will present current methods and applications on the modelling of bone tissue regeneration, with a focus on agent-based modeling approaches. I will show the application of these models to investigate the mechanisms behind successful and unsucessful regeneration as well for the development of treatment strategies.
Dez 6, 2023 - tba, NN
June 7, 2023. From mechanisms to prediction: Designing personalised treatment strategies for eczema. Reiko J. Tanaka.
Reiko J. Tanaka, Imperial College London
From mechanisms to prediction: Designing personalised treatment strategies for eczema
Atopic dermatitis (or eczema, AD) is the most common inflammatory skin disease characterised by inflamed, dry and itchy skin leading to substantial quality of life impairment and significant socioeconomic impact. Designing personalised treatment strategies for AD is challenging, given the apparent unpredictability and large variation in AD symptoms and treatment responses within and across individuals. A first step toward developing personalised treatment strategies is to better predict the consequences of possible treatments at an individual level, rather than at population level, to deal with the variability across patients.
We developed a mechanistic model of AD pathogenesis which provided a coherent mechanistic explanation of the dynamic onset, progression, and prevention of AD, as a result of interactions between skin barrier, immune responses and environmental stressors. Model predictive control using the mechanistic model suggested a possibility for designing personalized treatment strategies. We also adapted the structure of the mechanistic model to real patient data and developed a Bayesian mechanistic model tailored to each individual, that can predict the next day’s AD severity score given their score and treatments used on that day.
I would like to discuss hopes and challenges in designing personalized treatment strategies, for which detailed dynamic data is required but not often available.
May 10, 2023. 3 pm. Scaling up mechanistic modeling of disease processes using automated knowledge assembly. Benjamin Gyori
Benjamin Gyori, Director of Machine-assisted Modeling and Analysis. Harvard Medical School
Title: Scaling up mechanistic modeling of disease processes using automated knowledge assembly
Computational models of biological mechanisms take substantial human effort to construct and rarely scale to the level of omics datasets, while statistical approaches often do not make use of prior knowledge about mechanisms. At the same time, scientific knowledge is rapidly evolving with about 1.4 million new publications in biomedicine per year. To address these challenges, I present INDRA, an automated knowledge assembly system that integrates multiple text-mining approaches to process scientific literature combined with human-curated databases. INDRA standardizes knowledge extracted from these sources, systematically corrects errors, resolves redundancies, infers missing information, and introduces explicit probability models and machine-learned estimators of confidence to create a coherent knowledge base, up to the scale of all available literature. Detailed simulation models, causal networks, and knowledge graphs are generated from this assembled knowledge to support further analysis. I will present applications of this technology to explaining cancer gene co-dependency data and constructing explanations for experimental observations of drug response in multiple disease areas. This opens up directions toward increasingly automating the scientific discovery cycle with the help of machines building and using models and recognizing surprising observations or proposing new hypotheses.
*April 5, 2023. Dynamic logic models complement machine learning for personalized medicine. Julio Saez-Rodriguez. >>>including video<<<
Julio Saez-Rodriguez, University of Heidelberg
Location: online ZOOM - see video below
Dynamic logic models complement machine learning for personalized medicine
Multi-omics technologies, and in particular those with single-cell and spatial resolution, provide unique opportunities to study the deregulation of intra- and inter-cellular signaling processes in disease. I will present recent methods and applications from our group toward this aim, focusing on computational approaches that combine data with biological knowledge within statistical and machine learning methods. This combination allows us to increase both the statistical power of our analyses and the mechanistic interpretability of the results. These approaches allow us to identify key processes, that can be in turn studied in detailed with dynamic mechanistic models. I will then present how cell-specific logic models, trained with measurements upon perturbations, can provides new biomarkers and treatment opportunities. Finally, I will show how, using novel microfluidics-based technologies, this approach can also be applied directly to biopsies, allowing to build mechanistic models for individual cancer patients, and use these models to prose new therapies.
*February 1, 2023. Mechanistic insights into sensitization and desensitization of the Interferon α signal transduction pathway. Marcus Rosenblatt, Jens Timmer >>>including video <<<
Marcus Rosenblatt, Jens Timmer, Institute of Physics, University of Freiburg
Location: online ZOOM - see video below
Mechanistic insights into sensitization and desensitization of the Interferon α signal transduction pathway
As a key component of the innate immune system, Interferon alpha (IFNα) orchestrates the antiviral response in hepatocytes. The IFNα signal transduction pathway is known to desensitize upon activation which constitutes a major problem for the usage of IFNα as treatment against chronic viral infections or as an anti-tumor drug. However, the mechanisms that lead to this desensitization remain poorly understood.
Here, an ODE model is presented that describes the biochemical reaction network of IFNα signaling in different hepatoma cell lines as well as primary human hepatocytes (PHH). The calibrated model shows that besides a dose-dependent desensitization mediated by the negative feedback components SOCS1 and USP18 that act at the receptor level, the signaling pathway can also show (hyper-)sensitization in consequence of an upregulation of the intra-cellular components IRF9 and STAT2.
The model predicted the dose-dependent dynamics of transcriptionally active complexes in the signaling pathway and their effect on mRNA production, as shown by independent validation experiments. Furthermore, the model-based analysis of measurement data from PHH unraveled that each cell system establishes a particular dose-depending sensitization behavior whose shape is strongly determined by the abundance of the feedback components USP18 and STAT2. We show that our findings help to understand the dynamics of production of Interferon Stimulated Genes (ISGs) which exert numerous antiviral effector functions, and serve as a basis for a patient-individual optimization of the antiviral response upon IFNα stimulation.
*January 11, 2023. Modeling and Inference for Biological Systems with Hidden Components. Jae Kyoung Kim. >>> including video<<<
Jae Kyoung Kim, Dept. of Mathematical Sciences, KAIST, South Korea
Location: ZOOM (Video below)
Modeling and Inference for Biological Systems with Hidden Components
Despite dramatic advances in experimental techniques, many facets of intracellular dynamics remain hidden or can be measured only indirectly. In this talk, I will describe strategies to develop mathematical models with hidden parts: replacement of hidden components with either time delay or quasi-steady-state. Then, I will illustrate how the simplification with the time delay can be used to understand the processes of protein synthesis, which involves multiple steps such as transcription, translation, folding, and maturation, but typically whose intermediates proteins cannot be measured. Furthermore, I will illustrate how the simplification with the quasi-steady-state can be used to develop an accurate method to estimate drug clearance, which occurs in multiple steps of metabolism, which greatly improved the canonical approach used in more than 65,000 published papers for last 30 years. Finally, I will describe an inference method, GOBI (General Model-based Inference), that identifies hidden regulatory biochemical connections from time-series data. This method adopts the advantage of model-free inference methods (broad applicability) and model-based inference methods (accuracy).
*December 7, 2022. Semi-quantitative modeling in systems biology: the Petri net formalism. Ina Koch. >>>including slides<<<
Ina Koch, University Frankfurt
Semi-quantitative modeling in systems biology: the Petri net formalism
Modeling disease pathways requires the integration of various data of different type and quality. Whereas the data needed for deterministic approaches, such as ODE-based modeling, are often rare or not available, a huge amount of qualitative data have been generated by high-throughput methods in the last years. This causes the need for developing and applying new modeling approaches. Besides Boolean modeling, semi-quantitative modeling using Petri nets has been successfully applied to model biochemical systems, covering metabolic systems, signal transduction pathways, gene-regulatory pathways and hybrid systems.
In the talk, we will first introduce the basics of the Petri net formalism and its static and dynamic analysis techniques. The techniques range from basic graph-theoretical considerations and integer linear programming techniques to dynamic simulations. We will focus on invariant-based analysis approaches, in particular on transition invariants and Manatee invariants. We will illustrate the introduced terms using a case study of signal transduction in the NF-kB pathway before considering an exhaustive Petri net model of the TNFR1-induced signaling pathway, which covers cell survival, apoptosis and necroptosis processes. We will present the main results and additionally discuss the results of in silico knockout analysis based on the Petri net model.
November 9, 2022. Models of metabolism for biomedical research. Ralf Steuer.
Ralf Steuer, Institute for Theoretical Biology, Humboldt University Berlin
Models of metabolism for biomedical research
Cellular metabolism is the ultimate driving force of life. Metabolism provides the chemical energy required to run cellular processes, ensures cellular functions and removal of byproducts, and synthesises the building blocks for growing cells. Metabolic processes and the resulting metabolite levels can be interpreted as "end products" of cellular regulatory processes, and metabolic dysfunctions are the underlying cause of a multitude of diseases, including cancer.
The talk aims to discuss the current dichotomy between small-scale kinetic models, and comprehensive genome-scale metabolic reconstructions of metabolism in the context of biochemical research. In particular, while genome-scale reconstructions offer unprecedented possibilities to study the organization of human metabolism, their analysis is typically restricted to stoichiometric properties and stationary fluxes. In contrast, kinetic models of metabolism based on ordinary differential equations allow for a dynamic description, but their reconstruction requires extensive (and currently often not available) knowledge about enzyme-kinetic parameters. The talk will explore intermediate methods that (a) seek to derive smaller coarse-grained models from genome-scale reconstructions, and (b) use Monte-Carlo methods to account for unknown or uncertain parameters.
June 15, 2022. How mathematical models support clinical drug development: an industry perspective. Bernhard Steiert
Dr. Bernhard Steiert, Roche Pharmaceutical Research and Early Development, Roche Innovation Center Basel, Basel, Switzerland
How mathematical models support clinical drug development: an industry perspective
Mathematical modelling is a key component of drug development and is applied at all stages of a molecule progressing along the value chain. In clinical development, it is essential to find the right dose for every patient, which is composed of the right amount of drug and the right treatment interval for the right patient population. Modelling is also used to leverage and extrapolate the available data, informing decisions on which programs to accelerate and which ones to stop.
In this talk, an overview of the Roche Pharma Research and Early Development (pRED) setup for modelling will be provided. Typical models used in the pharmaceutical industry will be exemplified and, based on those, more general modelling themes are outlined and discussed. Involvement of key internal and external stakeholders is crucial for impacting decisions with modelling. I will summarize who they are and which communication strategies are most suitable for building trust in modelers and their models. The talk will close with examples where modelling was impactful and helped to bring safe and efficacious medicines to patients faster.
*May 4, 2022. Easy implementation and acceleration of spatial agent-based modelling of biological processes using rule-based modelling. Gavin Fullstone, Markus Morrison. >>>including video<<<
Dr. Gavin Fullstone, University of Stuttgart, Stuttgart Research Centre Systems Biology
Prof. Dr. Markus Morrison, University of Stuttgart, Institute of Cell Biology and Immunology
Easy implementation and acceleration of spatial agent-based modelling of biological processes using rule-based modelling
Agent-based modelling is particularly adept at modelling complex features of cell signalling pathways, where heterogeneity, stochastic and spatial effects are important, thus increasing our understanding of decision processes in biology in such scenarios. However, agent-based modelling often is computationally prohibitive to implement. Parallel computing, either on central processing units (CPUs) or graphical processing units (GPUs), can provide a means to improve computational feasibility of agent- based applications but generally requires specialist coding knowledge and extensive optimisation. In this seminar, we will demonstrate how we implemented and utilise a rule-based modelling approach to define particle-based models that can be flexibly parallelised on either CPU or GPU-architecture. In comparison with an established particle-based simulator, we achieve speed ups of >650-fold on modern GPUs whilst maintaining robust simulation. Additionally, we will demonstrate how we apply stochastic hybrid approaches to model the activity of multi-protein complexes and their contribution to cell signalling outcomes. In introducing these approaches, we will simultaneously present specific examples where we have successfully applied these approaches to help us understand fundamental biological processes, predict outcomes in biologically relevant scenarios and in the design of targeted therapeutics.
*April 6, 2022. Multi-scale modelling. Martin Falcke, Lutz Brusch. >>>including video<<<
Dr. Martin Falcke, Max-Delbrück-Centrum für Molekulare Medizin (MDC)
Dr. Lutz Brusch, Research Group Leader,Center for Information Services and High Performance Computing (ZIH), Technische Universität Dresden
Biological systems comprise many structural levels represented by molecules, cell organelles, cells, tissues, organs, organisms, populations and ecological levels. Each structural levels has its typical time and length scale from fs and nm on molecular level to km and years on ecological level. Multiscale modelling refers to mathematical models describing dynamics on different structural levels. Here, several structural levels and several scales may be accounted for either by a single equation or by multiple coupled equations each comprising only one scale. We present examples of both approaches.
In the first case, we present reaction diffusion systems for cardiac myocytes taking behaviour of individual ion channels into account and simulating whole cells at the same time. The mathematical problems and methods addressing them will be explained. For the second case, we introduce coupling of processes between intracellular, cellular and tissue scales. Such coupling requires that parameters of model processes on one scale get driven by the dynamical variables of model components at another scale. We demonstrate such model integration across multiple scales in the software framework Morpheus (https://morpheus.gitlab.io) and discuss the emergence of complex 'morphodynamic' behaviour that cannot be addressed by simpler models at a single scale.
*March 2, 2022. Response-Time Modelling of Cell-Cell Interaction Dynamics . Kevin Thurley >>>including video<<<
Prof. Kevin Thurley, Biomathematik, Institut für Experimentelle Onkologie, Universitätsklinikum Bonn, https://www.thurleylab.org
Response-Time Modelling of Cell-Cell Interaction Dynamics
Cell-to-cell communication networks have critical roles in coordinating diverse organismal processes, such as tissue development or immune cell response. However, compared with intracellular signal transduction networks, the function and engineering principles of cell-to-cell communication networks are far less understood. Major complications include: cells are themselves regulated by complex intracellular signaling networks; individual cells are heterogeneous; and output of any one cell can recursively become an additional input signal to other cells. In this talk, I will introduce a framework that treats intracellular signal transduction networks as "black boxes" with characterized input-to-output response relationships (Thurley K, Wu LF, Altschuler SJ, Cell Systems 2018, DOI: 10.1016/j.cels.2018.01.016). The approach can be used to predict communication network structure using experimentally accessible input-to-output measurements and without detailed knowledge of intermediate steps. Here, response-time modeling will be presented in the context of data-driven analysis of T cell differentiation and expansion in chronic viral infections.
*February 2, 2022. Large-scale pathway models for biomedical research. Jan Hasenauer >>>including video<<<
Prof. Dr. Jan Hasenauer, LIMES Institute, Universität Bonn
Target audience: PhD and master students, postdocs, group leaders in the modelling field.
Large-scale pathway models for biomedical research.
Pathway models are powerful tools in modern life sciences. They are based on knowledge of individual biological pathways and can be integrated into large-scale models to capture crosstalk. In this seminar, I will briefly introduce the modeling approach and methods used to create large-scale models. I will then present a large-scale model of cancer signal transduction with thousands of biochemical species and reactions as an example application. Variants of this model have been used for multi-omics analyses in murine model systems and human cell lines. We have developed workflows for parameterization and analysis of these models. I hope to discuss these approaches and future directions.
January 12, 2022. Using Disease Maps in Biomedical Research. Olaf Wolkenhauer, Shailendra Gupta, Matti Hoch
Prof. Dr. Olaf Wolkenhauer, Dept of Systems Biology & Bioinformatics, University of Rostock
Dr. Shailendra Gupta, Systems Biology & Bioinformatics, University of Rostock
Matti Hoch, Systems Biology & Bioinformatics, University of Rostock
Location: online ZOOM
Using Disease Maps in Biomedical Research
Olaf Wolkenhauer, Shailendra Gupta, Matti Hoch (Systems Biology & Bioinformatics, University of Rostock)
Disease maps are interactive, web-based representations of the phenotypic, cellular, and molecular processes underlying complex diseases in the form of knowledgebase. Disease maps provide a platform to integrate heterogeneous clinical and experimental data for formulating hypotheses on diagnostic, prognostic, and/or therapeutic markers using systems biology and bioinformatics-based methods. In the seminar, we shall contrast the disease map approach with small-scale pathway modelling efforts and present examples from cancer systems biology and demoing the Atlas of Inflammation Resolution (AIR) (https://air.bio.informatik.uni-rostock.de) The AIR contains 40 manually curated submaps including > 6T elements each of which are manually annotated. In addition, the AIR is enriched with regulatory layers of microRNAs, lncRNAs, and transcription factors that form a large molecular interaction map (MIM) with >23,000 molecular interactions. We developed workflows that facilitate the integration and analyses of multi-omics data directly on the disease maps and enable users to perform in silico perturbation experiments for formulating data-driven hypotheses. In the seminar, we hope for a discussion in which we compare the “disease map approach” with other pathway and network modelling approaches.