Stratifying Data for Visualization in SimBiology | SimBiology Tutorials for QSP, PBPK, and PK/PD Modeling and Analysis - MATLAB
Video Player is loading.
Current Time 0:00
Duration 0:00
Loaded: 0%
Stream Type LIVE
Remaining Time 0:00
 
1x
  • Chapters
  • descriptions off, selected
  • captions off, selected
      Video length is 7:00

      Stratifying Data for Visualization in SimBiology | SimBiology Tutorials for QSP, PBPK, and PK/PD Modeling and Analysis

      From the series: SimBiology Tutorials for QSP, PBPK, and PK/PD Modeling and Analysis

      With the interactive Model Analyzer app, you can easily visualize experimental data and simulation results. In this video, you will see how to stratify data to visualize different model quantities and experimental data sets in multiple plots.

      Published: 20 Dec 2019

      In this video, we'll talk about slicing data for visualization in SimBiology. And for this, we'll use a model of the SGLT2 inhibitor that contains 16 species and about 15 on custom parameters. While all of them are important for the overall dynamics of that system, you might only be interested in a subset of them for visualization.

      For instance, after a single simulation, you might be interested in generating several plots to gather time courses that belong together. In this example, the plot from the top shows this time course as related to the PK. It displays the concentration of the inhibitor in the plasma and in the GI tract.

      The second plot shows us time courses related to the PD part of that model. It displays the glucose production and utilization. The last plot shows us the AUC of the urinary glucose excretion over a time period of 24 hours, which might be a good biomarker to take decisions based on simulations.

      This type of slicing and visualization is not only important for simulation data, but also for experimental data. For instance, you might want to slice your experimental data depending on covariate values, like on these plots. So let's go ahead and reproduce those plots together in SimBiology.

      We will start with the simulation data, and for this I'll open a program I just preconfigured focusing on simulation. So I can now go ahead and click Run to run the simulation and generate the default plot. By default all states that have been logged in that program will be displayed on a single plot. So potentially I could go back to that program, click on States to Log, and uncheck any of the states I would like to hide in that plot,

      Another option would be to go to the Property Editor of that plot and uncheck the responses I would like to hide. A third option would be to select the responses in that list, and I could perform multiple selection by pressing Ctrl and click right to delete the responses.

      The last option would be to start from a blank plot. For this I'll click on New Plot, go to the workspace of my program and our last run and the results, and there select the states I would like to visualize. So I can perform multiple selection by pressing Ctrl and drag and drop them on that plot to visualize the five time courses I wanted to display.

      So now I can go to the response list and select time courses that belong together. Click right and create a new set. This will display them in a new subplot. Can now go ahead and do the same for the AUC, and now I have the configuration I wanted.

      So I can change the name of those sets to give them a meaningful name. I can change, potentially, the color of any of the time course. I can also go to the Plot Settings and give a title to this plot or change any other settings, like the scale of the access or the grading, for instance. Once I'm happy with those plots, I can go back to the plot, click right to export this plot to a MATLAB figure that can now be saved to any file format I need for documentation or publication.

      So please remark that by default, the sets were sliced vertically and the responses by color. But I can change this with any of the styles that is here available. So I could slice in a grid, horizontally, vertically, by color, using line style or transparency. So for instance, I could use line style for the responses. This might be useful for publication in black and white.

      It's important to understand that those plots are now linked to my program. So let me first revert back to color, go back to my program and change any of the settings like the dosing of the inhibitor, and I click here on overlay results to see both front and keep the last results of the last simulation. Now, both plots-- plot one and plot two-- will be updated with the new results.

      So as you can see, there is now the possibility to slice by scenarios. Those scenarios are my both runs. Run with dosing run one and run without dosing, run two. While on the table at the top I can choose the style for slicing, at the bottom I can define the parameters or properties of the style. But of course, I can change the style of the slicing.

      So here, the scenarios are displayed in different line style, but I could also choose horizontal to get two different columns-- the first one from run one with dosing, and the second one for run two without dosing. The same type of slicing can be used for a parameter scan.

      For instance, if I run a parameter scan with two parameters, those two parameters will appear in the list of slicing properties. In this example, I chose a vertical style for the first parameter of the scanning, and horizontal for the second parameter of this scan. So now, I get the plot matrix with all parameter combinations that were used for scanning.

      This type of slicing is not only interesting for simulation data, but also for experimental data. I've here a data set that I will display in a data sheet first. This data set contains a grouping variable, it contains a time column, a dosing variable, it contains one column for measurement, and two covariates.

      So now I'll go ahead and create a plot for this data. So I click on New Plot to create a blank plot and drag and drop this data set on this plot. So by default, the scenarios will be the grouping variable, and they will be used for slicing using color. This is visible in the first table in the Property Editor.

      But as you can see now, I have also the other columns-- dosing columns and covariates-- as slicing properties available. So for instance, I might not be interested in having different groups in a different color. So I remove the color style for scenarios, and instead use this color for Apgar, my covariate.

      I could also use a style for weight. In this case, I chose grid to have different ridges of rates in a different plot. So since our covariates are continuous by nature, SimBiology will bend them into ranges. By default, the number of bends is set to four, but I can change this number.

      Here for weight, this will create, then, six subplots. So now please go ahead and discover the best way to visualize your data in SimBiology.

      Related Products