The ProBMoT release contains examples of modeling tasks from two domains.
The first domain comes from ecology and deals with modeling population dynamics in aquatic ecosystems. The second domain is in the field of systems biology and considers dynamical change of protein domain concentrations in the process of endocytosis.
The domains are briefly introduced and described in separate tab-sections (Aquatic and Endocytosis, respectively) below. Each description includes the files necessary to run ProBMoT for three specific tasks related to modeling in the corresponding domain: simulating a completely specified model of a dynamical system, estimating the values of model parameters from data, and inducing the model structure and parameters from data.
The structure of this library is complex, since it compiles an extensive set of model components used for modeling population dynamics in aquatic ecosystems. It specifies a hierarchy of key entities that emerge in aquatic ecosystems: primary producers (i.e., phytoplankton), zooplankton, nutrients and the environment they share. The observed interactions between these entities are: primary producer growth that is limited by the concentrations of nutrients and various environment factors (light and temperature, in particular), the mortality and respiration of primary producers and zooplankton being influenced by the environmental factors, and zooplantkon grazing on primary producers. For complete description of the library and all its components please refer to Cerepnalkoski et al. (2012).
The ProBMoT release includes this library in the file examples/aquatic/AquaticEcosystem.pbl
The data sets are available at this location.
The ProBMoT release includes these data in the folder examples/endocytosis/data/.
The data sets comprise daily measurements of a number of observed variables (concentrations of phyto- and zoo-plankton, nitrogen-, phosporus- and silica-based nutrients, light intensity and temperature) in Lake Bled, Slovenia in the period from 1996 to 2002.
This process-based model represents a completely specified model (both in terms of its structure and parameters) of the phytoplankton dynamics observed in Lake Bled. It has been induced with ProBMoT from the data measured in 1996.
The ProBMoT release includes this model in the file examples/aquatic/BledComplete.pbm.
The accompanying task specification allows for simulation of the particular model. Note that for its simulation, one have to specify a data set that includes measurements of the exogenous (input) variables of the model. Note the selected ProBMoT command simulate_model and the missing setting for the fitter of the parameter values, which is not needed for simulation.
The ProBMoT release includes this task specification in the file examples/aquatic/taskSimulate.xml.
After running ProBMoT on this particular task (using the command java -jar probmot-1.2.jar examples/aquatic/taskSimulate.xml) the output files are written in the folder examples/aquatic/out/simulation/ (set in the task specification file).
The second incomplete process-based model specifies a fixed single model structure and leaves the values of the model parameters unspecified. It reuses the structure of the complete model used above for the simulation task.
The ProBMoT release includes this model in the file examples/aquatic/BledIncompleteParameters.pbm.
The accompanying task specification is used to perform parameter estimation task with ProBMoT: given a model structure with unknown parameters and measured behavior, find model parameter values that lead to simulated behavior that closely fit the measured one. Note the selected ProBMoT command exhaustive_search: in the particular case, no search through the space of model structures is performed since the incomplete model specifies a single model structure.
The ProBMoT release includes this task specification in the file examples/aquatic/taskEstimate.xml.
After running ProBMoT on this particular task (using the command java -jar probmot-1.2.jar examples/aquatic/taskEstimate.xml) the output files are written in the folder examples/aquatic/out/estimation/ (set in the task specification file).
The third incomplete process-based model leaves both the structure and the values of the model parameters unspecified. A set of constraints is imposed on the structure space by instantiation of processes from from a higher level template processes defined in the library.
The ProBMoT release includes this model in the file examples/aquatic/BledIncompleteStructure.pbm.
The accompanying task specification is used to perform structural identification task with ProBMoT: given a set of model structures with unknown parameters and measured behavior, find a model and corresponding model parameter values that lead to simulated behavior that closely fit the measured one. The selected ProBMoT command is exhaustive_search.
The ProBMoT release includes this task specification in the file examples/aquatic/taskInduce.xml.
After running ProBMoT on this particular task (using the command java -jar probmot-1.2.jar examples/aquatic/taskInduce.xml) the output files are written in the folder examples/aquatic/out/induction/ (set in the task specification file).
The structure of the library is based on a modular formulation of the system of differential equations for modeling the conversion between the Rab5 and Rab7 protein domains as presented in the study by del Conte-Zerial et al. (2008). It encodes a single template entity, presented that refers to a general protein domain. In also encodes, using hierarchical template processes, all possiblities for the different possible interactions between the protein domains, conforming to the aforementioned modular formulation. More details can be found in Tanevski et al. (2015).
The ProBMoT release includes this library in the file examples/endocytosis/EndocytosisLibrary.pbl.
The data set is available at this location.
The ProBMoT release includes this data set in the file examples/endocytosis/data/endo.data.
The data used for the task of modeling the Rab5-Rab7 switch in endocytosis is an aggregation of the experimental data obtained by the group of Marino Zerial at the Max-Planck Institute for Cell Biology and Genetics in Dresden, Germany. It contains the total density of Rab5 and Rab7 proteins in 10,571 time points along the time interval of [823.56, 1128.56] seconds which corresponds to a time window in which a switching event is observed.
This process-based model represents a completely specified model with fixed structure and parameters. Its structure and parameter values correspond to the best performing structure and parameter values that can achieve Cut-Out/Toggle switch like behavior found in the study by Tanevski et al. (2015).The parameters with values set to 0 are not used by the specific process.
The ProBMoT release includes this model in the file examples/endocytosis/EndocytosisModelCOT.pbm.
The accompanying task simulates the complete model that has a structure able to achieve Cut-Out/Toggle switch like behavior. In this task specification the path to the complete model is the value of a <model> element denoting that a completely specified (structure and parameters) model is given on input; The command given to ProBMoT is simulate_model and the settings for the fitter and the initial values are omitted.
The ProBMoT release includes this task specification in the file examples/endocytosis/EndocytosisTaskSimulateCOT.xml.
The output of the task will be stored in the folder examples/endocytosis/out/EndocytosisModelCOT/ after running ProBMoT with the command java -jar probmot-1.2.jar examples/endocytosis/EndocytosisTaskSimulateCOT.xml.
This incomplete model defines a set of constraints for instantiation of processes that identify only a single possible structure. It instatiates all possible processes from corresponding template process defined at the lowest level of hierarchy of templates. This structure corresponds to the best performing structure that can achieve In-Phase switch like behavior found in the study by Tanevski et al. (2015).
The ProBMoT release includes this incomplete model in the file examples/endocytosis/EndocytosisIncompleteIP.pbm.
The task uses the incomplete model with fixed structure which can achieve In-Phase switch like behavior as an input. Although the command given to ProBMoT is exhaustive_search the definition of only a single model structure in the incomplete model reduces this task from structural identification to the task of parameter estimation.
The ProBMoT release includes this task specification in the file examples/endocytosis/EndocytosisTaskFitIP.xml.
The output of the task will be stored in the folder examples/endocytosis/out/EndocytosisIncompleteIP/ after running ProBMoT with the command java -jar probmot-1.2.jar examples/endocytosis/EndocytosisTaskFitIP.xml.
Given the library of domain knowledge, this incomplete model defines all possible structures using a most general set of constraints. It instatiates each speciffic process from the corresponding top-level template process allowing for every possible combination of the defined alternatives. Using this incomplete model a total of 126 model structures can be enumerated.
The ProBMoT release includes this incomplete model in the file examples/endocytosis/EndocytosisModel.pbm.
The task specification for performing exhaustive_search corresponds to the best performing task used in the article by Tanevski et al. (2015). It exhausitvely enumerates and fits 126 models to the available data using the ERX objective function with parameter alpha=0.5.
The ProBMoT release includes this task specification in the file examples/endocytosis/EndocytosisTask.xml.
The output of the task will be stored in the folder examples/endocytosis/out/EndocytosisModel/ after running ProBMoT with the command java -jar probmot-1.2.jar examples/endocytosis/EndocytosisTask.xml.