Information on Black Board Teaching and Computer Practicals

Please note that you will be able to visit one Blackboard Teaching Session (BB) or Computer Practical (CP) per day, which will take place on Thursday, Friday and Sunday afternoon from 16:45 to 18:45. Please chose one topic per day of the selection listed below. All in all you have 10 different topics to choose from. Please note, that all Computer Practicals and Blackboard Teachings will be held twice, except BB04 & BB05.

Last update: 28.02.2018: BB6: Alberto Marin-Sanguino substituted for Michael Savageau

Overview - Blackboard Teaching and Computer Practicals per day

Registration for the Blackboard Teaching and Computer Practicals

The registration for the Blackboard and Computer Practicals can be found in the User Profile (login required). The number of places for these courses are limited and will be assigned on a first come, first served basis.

For computer practicals, students are required to bring their own laptops with any required software preinstalled (preinstallation requirements listed at the end of each course).

BB01 "Stochasticity"

Frank Bruggemann

TBA

BB02 Constraint-based modeling: from Flux Balance Analysis to resource allocation modeling

Bas Teusink (Amsterdam, NL)

From a sequenced genome, the reconstruction of the metabolic network of an organism is the fastest path to increased understanding of its physiology (see ref [1] for a recent example). Such a reconstruction is an inventory of all metabolic reactions that an organism can potentially carry out – provided the bioinformatic inference of gene function was correct and the corresponding protein is expressed and active [2].
In this blackboard session, I will explain the basics of the reconstruction process, and then discuss the uses and limitations of popular constraint-based modeling techniques applied to the metabolic network, Flux Balance Analysis in particular. I will then explain the concept of resource allocation, and show how this can be incorporated into the constraint-based modeling format to improve model predictions. Depending on the interest of the audience, and time left, we can then continue the discussion to different applications or extensions of the approach.

1. Branco Dos Santos F, Olivier BG, Boele J, Smessaert V, De Rop P, Krumpochova P, et al. Probing the genome-scale metabolic landscape of Bordetella pertussis, the causative agent of whooping cough. Appl. Environ. Microbiol. 2017;
2. Santos F, Boele J, Teusink B. A practical guide to genome-scale metabolic models and their analysis. Methods Enzymol. 2011;500:509–32.

BB03 "Enzyme regulation, thermodynamics"

Wolfram Liebermeister

TBA

BB04 Modeling spatio-temporal processes

Franziska Matthaeus (Frankfurt, DE)

Biological components do not only interact, they are also subject to transport processes, such as diffusion or drift. Transport processes influence reaction kinetics and lead to spatiotemporal phenomena like traveling waves or the formation of spatial patterns. A common approach to model spatiotemporal processes are partial differential equations. We will derive, analyse and discuss classical descriptions of transport and reaction-diffusion equations. The course will cover diffusion, including properties of diffusion and the derivation of the diffusion equation, meaning and estimation of the diffusion coefficient, advection, anomalous diffusion, chemotaxis, and numerical approaches to solving the respective PDE’s. We furthermore consider reaction-diffusion systems and derive conditions for traveling waves and spatial patterns. The model system for traveling waves, the Fisher-Kolmogorov equation, couples diffusion with logistic growth and exhibits solutions of fronts moving with constant profile and speed. In case of spatial patterning, we consider Turing systems, coupled reaction-diffusion-equations, with distinct diffusion coefficients and specific nonlinear reaction terms. Apart from the mathematical background we will present many biological examples where these processes are of importance.

BB05 Metabolic pathway design and analysis

Elad Noor

In this blackboard session, we will cover several aspects of pathway design in the context of metabolic engineering:
- Stoichiometric modeling and flux coupling algorithms
- Thermodynamic constraints
- Directed evolution

BB06 A Novel Strategy to Accelerate the Modeling and Analysis of Complex Biological Systems

Alberto Marin-Sanguino (substituted for Michael Savageau)

Although we have a well-defined concept of an organism’s genotype, its phenotypes – the biological functions implemented by its underlying biochemistry – are difficult to define and predict. We ultimately want to represent the phenotype by a mechanistic model that accurately describes the changes in the concentrations of the compounds under various conditions. The ‘architecture’ of a system can be inferred from high-throughput data, but numerous unknown kinetic parameters influence exactly how change in one concentration affects others. Often the phenotype becomes clear only when those parameters are known; even a simple model can exhibit many phenotypes given different parameter choices. Current approaches to determining phenotype thus focus first on finding parameter values for the underlying biochemistry, typically through a mixture of ad-hoc experimentation and computationally inefficient high-dimensional numerical search. While these strategies have been used to fully characterize small systems in the pre-genomic era, a mechanistic understanding of systems, even of moderate size, derived from genotype data remains elusive. We propose a fundamental shift towards a post-genomic computational paradigm in which we first analytically determine the space of possible phenotypes for a given network architecture and then predict parameter values for their realization, predictions that can guide experimentation and further numerical analysis. This ‘phenotype-centric’ paradigm combines four innovations with the potential to accelerate our understanding of complex biological systems: (1) a rigorous mathematical definition of biochemical phenotypes, (2) a method for enumerating the phenotypic repertoire based on the biomolecular network architecture, (3) an integrated suite of computational algorithms for the efficient prediction of parameter values and analysis of the phenotypic repertoire, and (4) a user-focused environment for navigating the resulting space of phenotypes and identifying biologically relevant features. These innovations will facilitate deterministic and stochastic simulations that require parameter values, will accelerate both hypothesis discrimination in systems biology and the design cycle in synthetic biology, and will enable investigators to achieve predictive understanding of biomolecular phenotypes from genotype. I will describe recent progress toward the realization of this potential and some of the remaining challenges.

CP01 Advanced topics in parameter estimation with COPASI

Pedro Mendes (University of Connecticut, US) and Sven Sahle (University of Heidelberg, DE)

COPASI is a widely used tool for creating, simulating and analysing kinetic models of biochemical reaction networks. In this session we will focus on parameter estimation/model fitting, which is one of the key features of COPASI and at the same time a crucial technique in computational systems biology. We will have practical demonstrations and exercises about setting up and running parameter estimation tasks with the software COPASI, including setups using data from multiple experiments under different conditions. We will discuss practical issues including the choice of optimisation algorithms (local/global optimisers). We will then cover how model fitting is used to answer qualitative and quantitative biological questions. (E.g. "Can my hypothesis explain the observed experimental data?“ and "Can I resolve the quantitative properties of biochemical processes from the given data?“). We will show how parameter identifiability can be analysed using COPASI. During the week of the course we will be available to answer questions on the features of COPASI that will not be covered in this session.

Software: Please download and preinstall Copasi (latest stable version) for this practical.

CP02 CellDesigner: A process diagram editor for gene-regulatory and biochemical networks

Akira Funahashi (Keio, JP)

CellDesigner is a software package for modeling and simulation of biochemical and gene regulatory networks, originally developed by the Systems Biology Institute in Japan. While CellDesigner itself is a sophisticated structured diagram editor, it enables users to directly integrate various tools, such as built-in SBML ODE Solver, COPASI, Simulation Core Libray and SBW-powered simulation/analysis modules. CellDesigner runs on various platforms such as Windows, MacOS X and Linux, and is freely available from our website at http://celldesigner.org/. In this course, I will explain how CellDesigner can be used from modeling perspectives. The first topic will feature network modeling using CellDesigner, and will show how she/he could build a model from scratch, and examine simulations. This topic also includes an explanation on how we build a biochemical network as a "map" which includes links to several existing databases, and how we build a mathematical model by the aspect of process-diagram based modeling. Once a model is described with appropriate mathematical equations and parameters, running a simulation on CellDesigner is quite straight forward. I will also explain how to easily tweak your model from CellDesigner's user-interface and observe some changes in the dynamics. Not just building a model from scratch, this course also introduces how we can "import" an existing model from several third-party databases (ex. BioModels.net, PANTHER database). This might be useful for users who actually read a paper and got interested in the model, but does not have enough experience on building a mathematical model by hand. This tutorial will cover both mathematical modeling and graphical modeling (create a model as a 'map') topics, and mainly focuses on mathematical modeling. Bringing your notebook PC with CellDesigner installed is highly recommended.

Software: Please download and preinstall CellDesigner for this practical.

CP03 VCell: compartmental and spatial modeling and simulation using reactions and reaction rules

Michael Blinov

Virtual Cell (VCell, http://vcell.org) is an open-source platform (automatic installers for Windows, Mac OS and Linux) that provides powerful capabilities for kinetic modeling of cellular systems. It provides one-stop simulation shopping: deterministic (compartmental ODE or reaction-diffusion-advection PDE), stochastic reactions (several SSA solvers), spatial stochastic (reaction-diffusion with Smoldyn), spatial hybrid deterministic/stochastic and compartmental network-free agent based simulations.
Models can be formulated as an explicit network or generated from graphically expressed rules. Modelers can enter reactions and pathways in a biology-based interface, VCell automatically creates the math for you. Optionally modelers may also enter math directly. VCell supports multiple biophysical mechanisms, including reaction kinetics, diffusion, flow, membrane transport, lateral membrane diffusion and electrophysiology.
Models can be saved on a central searchable database and simulations can run on our remote servers from any low-cost laptop. Models and their simulation results can be accessed from anywhere, and models can be shared among collaborators or made publicly available. Optionally, simulations can also be run locally without an Internet connection.
An important feature of VCell is to allow modelers to ask how spatial features of cells affect the system behavior. At the simplest level, the relative sizes of compartments affect the concentrations of species transported between them. If diffusion and spatial localization of molecular species can affect the biology, the geometric shapes of the membrane and volumetric compartments can be explicitly considered. Model geometries may be derived from idealized analytical expressions or from experimental 2D or 3D microscope images, or constructed using constructive solid geometry.
VCell supports modeling using a combination of reaction networks and reaction rules (rule-based modeling). It provides a way to account for all the potential molecular complexes and interactions among multivalent or multistate molecules. Molecular interactions can be specified in the form of reaction rules that serve as generators of reactions. The same set of rules can be simulated using both network generation (BioNetGen software) and network-free simulation (NFSim software).

Software: Please download and preinstall Virtual Cell (Release version VCell 6.1) for this practical.

CP04 Build better models by turning redundancy into actual modularity with Tellurium - a python modeling environment

Tom Altenburg (Berlin, DE)

This computer practical illustrates possible applications of Tellurium, which is a Python environment for building, solving and analyzing models in systems biology. The course will begin with a short introduction to i) antimony, which is Tellurium’s human-readable modeling language and ii) roadrunner, Tellurium’s solver intended to simulate those models. From my point of view, Tellurium reflects an optimal mixture between simplicity and controllabilty for both the model and the workflow. Additionally, the modeling language antimony was specifically invented to be able to instanciate a model. A feature which supports building complex models in a modular manner.
Hence, applications of modularity while implementing antimony models will be shown for specific scenarious in this course. The aspect of modularity will be discussed in case of technical or biological redundancy possibly occuring in a system. We will focus on antimony’s modular syntax allowing the re-usage of those redundant parts, which typically results in an overall more readable and lightweight model. Apart from that, a complex model may be divided into complementary sub-models, which can range from distinct pathways down to functional place-holders. We will investigate how to integrate those parts and how to implement connections between modules in antimony.
Overall the concept of modular modeling provides some major advantages as sub-models can be checked independently, exchanged for hypothesis testing or updated to include the latest advances of a field. The application of those tasks are motivated and illustrated in this course trying to point out a typical workflow in Tellurim’s environment. Situations that are incompatible with a modular approach are also indicated to the students.
For this course it is useful to have a basic experience with coding in general and modeling in systems biology.
IMPORTANT NOTE:
Tellurium, Python and Jupyter must be installed on your laptop BEFORE the practical starts. Recommended is the usage Anaconda and Pip to install all three packages (works for Linux, Mac and Windows). If you are unfamiliar with those installation tools, please ask for help via mail: tom.altenburg@hu-berlin.de

Software: Tellurium, Python and Jupyter must be installed on your laptop BEFORE the practical starts. Recommended is the usage Anaconda and Pip to install all three packages (works for Linux, Mac and Windows). If you are unfamiliar with those installation tools, please ask for help via mail: tom.altenburg@hu-berlin.de