Computational modelling of cellular processes: regulatory vs metabolic systems
Tuesday, 1st September
13:30 - 16:30 (CEST)
Instructors and helpers
- Anna Niarakis | Université d'Évry, Université de Paris-Saclay, France
- Dagmar Waltemath | University Medicine, Greifswald, Germany
- Pedro Monteiro | Universidade de Lisboa, Portugal
- Vincent Noel | Institut Curie, France
- Marta Cascante | Universitat de Barcelona, Spain
- Miguel Ponce de Leon | Barcelona Supercomputing Center, Spain
Motivation: The goal of this tutorial is to provide hands-on experience in modelling software tools, which are well suited to study different types of cellular processes. In particular, this tutorial aims to present approaches used to study models of regulatory versus metabolic systems, which rely on distinct formalisms and are rarely presented together.
Logical modelling provides a computational approach for the visualization and analysis of the dynamics of biochemical and biological systems. One of the main advantages of logical models is their scalability and the relatively easy method of construction. In part due to these attributes, logical models have become increasingly more popular among the computational biology community. This has, in turn, led to the development of different techniques and software tools that enable the construction, simulation, and analysis of logical models and their variants (Boolean, multilevel, deterministic, stochastic, etc.) to address various biological questions.
Petri nets are executable graphs that can be used to model the flow of information (tokens) through a system. They are composed of two types of nodes: places, which in a biological context can be used to represent entities such as proteins or metabolites; and transitions, which can be used to represent reactions between them. Directional edges are used to connect the two types of nodes. These concepts are closely aligned to that of process diagrams, such that when formerly drawn in this manner and parameterised, graphical models can be used to simulate ruled-based activity flow through complex models of biological systems.
Constraint-based modelling has been successfully used to model metabolic networks at genome-scale, in particular it is the basis of the widely used FBA method and its derivatives (pFBA, MOMA, ROOM). Constraint-based methods mainly rely on the use of the stoichiometry of the reactions and it used physicochemical constraints, such as mass conservation, to define the space of feasible metabolic states. While logical models usually predict coarse grain cellular phenotypes (e.g. proliferation or apoptosis), constraint-based method aims to predict the metabolic state of the cell in terms of flux values. This allows the predictions of growth rates and import/excretion rates. Moreover, constraint-based methods also allow integrating omic dataset (e.g. gene expression) to generate context-specific models which can be used to study the metabolism of human cells in different conditions. Connections between regulatory and metabolic models is a required step in the roadmap to create whole-cell models. Herein, current approaches to bridge different formalism will be also discussed.
Presenting these approaches side by side will allow to highlight differences but also to identify interfaces for model integration and formalism coupling. The creation of multiscale, hybrid models would permit the study of intertwined cellular processes, like signalling, gene regulation and metabolism.
Scope: The tutorial will begin with a general introduction on the different levels of cellular processes and the methods most adequate to describe them. The tutorial will then be followed by four sessions on specific software tools described below. Each session will include a brief introduction on the biological question posed, the method of choice and the corresponding software, followed by interactive hands-on. The participants will thereby be trained on software use and application to treat a certain problem and most importantly, on how to define the inputs and interpret the outputs in regards to the biological problem.
To help facilitate this objective, we will encourage participants to come prepared with their own research questions. The tentative schedule of the day covers the use of the following software tools:
- The CoLoMoTo notebook (https://colomoto.github.io/colomoto-docker/) is an interactive platform to build analysis workflows for qualitative models, integrating a collection of state of the art software tools in a common Python API.
- MaBoSS (https://maboss.curie.fr) is a stochastic simulator for continuous/discrete time Markov processes, applied on a Boolean network. It allows computing of time- dependant probabilities of the system's states, and can also use a specific language for associating transition rates to each node.
- OpenCOBRA (https://opencobra.github.io/) is a Python/Julia API to access COBRA (Constraint-based reconstruction and analysis) toolbox in order to provide a molecular mechanistic framework for integrative analysis of experimental molecular systems biology data and quantitative prediction of physicochemically and biochemically feasible phenotypic states.
This tutorial will be designed for participants (students and researchers) with no previous modelling experience, as well as to those who are seasoned modellers. Participants will be provided with a working docker or VirtualBox setup, later on the tutorial website.
This tutorial is open to at most 25 attendees.
No special requirements needed. Only a computer/laptop and interest in modelling.
|13:30 - 14:00||Building graphical and computational models in biology - Anna Niarakis|
|14:00 - 14:30||Model repositories and standard formats for model reusability - Dagmar Waltemath|
|14:30 - 15:00||Attractors and reachability analysis with the CoLoMoTo Notebook - Pedro Monteiro|
|15:00 - 15:30||Stochastic quantification of reachability with MaBoSS - Vincent Noel|
|15:30 - 16:30||OpenCOBRA - Marta Cascante / Miguel Ponce de Leon|