Modelling the cell 1 Flashcards
What models do we distinguish?
1) Descriptive (verbal)
2) Graphical (metabolic schemes)
3) 3-D physical (stick and ball models)
4) Mathematical (e.g. algebraic or differential equations)
1) Mechanistic vs. phenomenological
> Mechanistic: equations based on a mechanism: you know how something works (e.g. the way planets move).
Phenomenological: observations and try to describe them with equations. You don’t know what makes the system behave as it does.
2) Dynamic versus static
> Dynamic: time representation in system (plotting time on x-axis)
Static: equation describes the data (e.g. linear line)
3) Continuous time vs. discrete time models (dynamic)
> continuous: continuous representation
> Discrete: time steps (day 1, day 2 or year 1, year 2)
4) Spatially heterogeneous or homogeneous
> Hetero: concentration of metabolites ins not the same everywhere
Homogeneous: concentration of metabolites is the same everywhere in the cell
4) Spatially heterogeneous or homogeneous
> Hetero: concentration of metabolites ins not the same everywhere
Homogeneous: concentration of metabolites is the same everywhere in the cell
5) Stochastic versus deterministic (parameters undergo random changes or are constant)
> Stochastic: change will determine the values
> Deterministic: parameters have a fixed value
What can you do with models?
- Understand the system that is being modeled (give a quantitative description of the system on the basis of the characteristics of the components)
- Make predictions of future states (or otherwise unknown states)
- Control the system to produce a certain output (by manipulation of parameters on the basis of model stimulations)
Methods in metabolic modeling
- Time hierarchies: separation of metabolism from gene expression
- Structural hierarchies: reaction to cell level, organism, population
- Ordinary differential equations (production-consumption etc)
- Enzymology (enzyme kinetic rate)
- Knowledge of reaction network structure up to cellular level
Types of models – levels of detail
1) Core models: as simple as possible (to test a hypothesis: this can happen)
2) Detailed kinetic models: simulating ‘reality’ (this is what happens)
Core model for glycolytic oscillations ? How do you get them?
must have at least two variables to get oscillations. Product induction or substrate inhibition
Validation of model: experiments. Why must certain tests be performed that are different from the experimental setup to build the model?
-> Building model construction: Fit parameters on a certain dataset to get a certain behavior. You need an independent data set to test how well the model functions. Can it predict the dataset of a new experiment? Then, you can trust the model more and more.
Bottom-up approach in making a model?
-> characterize components -> build rate equation -> parametrize rate equation -> set up set of differential equations to predict how the system will behave. It’s a prediction.
top-down model?
-> you get data points and fit your model to describe the data. (so on the basic of systemic data sets. Can I describe the set with the model?)
The model construction and validation workflow should be transparent and reproducible. How does one do that?
- All data should be made available, preferably in annotated format on a publicly accessible platform, e.g. EURO-SEEK
- Stick to standards, e.g. SBML, CellML
- Annotate models e.g. MIRIAM
- Store models in databases e.g. Biomodels or JWS
- Link to models and data in publications
