Tutorial

Welcome to the Cell Cycle Browser (CCB) tutorial. As you follow along the tutorial, you can check out the Glossary if you encounter terms you don't understand. The bullet points provide a quick-start to the tutorial. The text between each bullet point provides more in-depth information for understanding the finer features of the browser.

Section I: Visualizing the Cell Cycle

Navigate your browser to https://cellcycle.renci.org.

The CCB is divided into four panels:
- The Data Selection panel, stretching across the top of the browser, allows you to select a workspace consisting of a collection of live-cell datasets and cell-cycle models. You can also add/remove live-cell datasets, switch between models, and choose analysis plots to display.
- The Data Browser panel, in the center, displays one or more tracks of single-cell data and/or model simulation output. Time-series traces from individual cells are represented as a heat map in which the intensity of signal is denoted by a change in color (white: low, black: high).
- The Model panel, on the left-hand side, displays a graphical representation of the cell cycle based on a computational model, along with controls to adjust the model.
- The Analysis Plots panel, on the right-hand site, displays different analysis plots that provide additional information about the experimental and simulated data.
Hover over any information icon to receive pop-up information about that section of the CCB. Click the Save SVG button in any plot to download an SVG version of the plot.

Time-Series Analysis

Load the RPE-1 Demo Workspace by selecting it from the Workspace dropdown menu.

This dataset contains single-cell traces acquired from retinal pigment epithelial cells. The cells contain separate fluorescent biosensors for two biological species: 53BP1-mVenus and PCNA-mCherry. We use the term species to refer to a distinct pool of molecules that have either been measured (e.g., 53BP1-mVenus levels acquired by live imaging) or simulated by a computational model (e.g., 53BP1 produced through model simulation). In the Data Browser, each image feature for a species occupies a separate track. The feature is written in italics next to the reporter name (e.g. PCNA–Count) in the track header. In the current RPE-1 workspace, 53BP1 is represented by the feature foci # (a raw count of the number of 53BP1 foci) and PCNA is represented by std var (a measure of punctate patterning in the PCNA signal within the nucleus). Other features could include mean intensity or cell size.

Each track can be collapsed using the Collapse track button Collapse button . When a track is collapsed, the average time series remains visible.

Tracks can also be sorted using the different options in the Sort tracks bar, or arranged manually by clicking and dragging the header of any track and releasing it in the gap between any two other tracks.

Mouse over the shaded rectangles in the 53BP1-mVenus track to see raw values for the number of 53BP1 foci over time for an individual cell.

Time series for individual cells can be visualized in the Time Series plot in the Analysis Plots panel. Selected average traces for an entire track are shown with thicker lines, and traces for individual cell traces are shown with thinner lines.

Plot a time series of 53BP1 foci for an individual cell by clicking the Toggle time-series display button next to a row of single-cell data in the 53BP1-mVenus track. Note that time series of all features for that cell (i.e. in other tracks) can be turned on or off using the green check mark or red x buttons.
Mouseover a particular time series to highlight it. Note that the cell names in each track are also highlighted. You can also mouseover a trace name to highlight its time series (if visible).

Examine the traces for PCNA in the PCNA-mCherry track. Note that they have the same overall pattern: a stretch of light gray, followed by a stretch of gray intensity increasing to a dark peak, followed by another stretch of light gray.

PCNA Trace

This pattern corresponds to the cell cycle phases G1, S, and G2/M. Although the patterns are qualitatively similar, some traces show more intense gray values than other traces. It is often useful to rescale each trace individually so that the minimum and maximum gray values are the same. Such rescaling emphasizes the temporal behavior rather than the amplitude of the signals.

Rescale the traces for PCNA by clicking on the Rescale traces button .

Note that the gray values show a greater dynamic range and that traces look more similar.

It is often useful to align single-cell traces according to when the cell cycle begins or ends. The Data Browser enables you to left align or right align the single cell traces.

Align single-cell traces to the end of the cell cycle using the Align right button.

Notice that the dark gray values in the PCNA track representing the S/G2 transition are better aligned than when the traces were aligned to the beginning of the cell cycle. Note also that changing the alignment also updates the Time Series plot.

You can also stretch data so that all traces are aligned at both the beginning and end of the cell cycle.

Stretch single-cell traces to a common time axis using the Stretch data button.

Note that stretching the traces changes the x-axis from time units to % completion, since the time units have been distorted for individual cells.

It is often useful to be able to group similar traces together. To facilitate this, you can perform hierarchical clustering based on the average difference between traces at each sample point using the current rescale and alignment settings.

Perform hierarchical clustering by pressing the Cluster traces button .

A dendrogram is drawn next to the traces to show the hierarchical clustering. Note that the first trace is the most dissimilar, and is connected directly to the root. You can toggle dendrogram display with the Show dendrogram button , and sort by the original ordering with the Sort traces button .

Growth Curve Analysis

Let’s now examine the other visible analysis plot: the Growth Curve plot. Growth curves for each dataset are calculated from the average cell cycle duration using the formula: y = 2 ^ (t / duration). So we would expect that cells with a shorter cell cycle duration would proliferate faster and show a steeper growth curve. Let’s now compare the growth curve of RPE-1 cells to that of stem cells.

Load the H9/OCT4-mCherry live-cell data by selecting it from the Datasets dropdown menu.

Stars next to a dataset indicate datasets belonging to the current workspace.
Examine the Growth Curve plot in the Analysis Plots panel. Note that there are now two curves: one for RPE cells and one for H9 stem cells.

Note that the traces from RPE and H9 cells come from separate datasets, or sources. A data source constitutes a single live-cell experiment, or simulation. Typically cells from the same experiment are imaged in the same well or a nearby well. Cells from the same dataset may contain more than one biosensor and multiple features per biosensor. There is one growth curve per data source.

Cell Cycle Analysis (Phase Distributions)

For certain datasets, we can also perform a virtual flow cytometry cell cycle analysis. If information about cell cycle phases is known for individual cells, we can construct a scatter plot of the steady-state distribution of a large number of cells.

Load the FUCCI Live-Cell Data by selecting it from the Datasets dropdown menu.

Note that there is a phase track because phase information is included in the dataset as an image feature. Future versions of the CCB may include the ability to estimate phase locations based on single cell data.
Examine the Cell Cycle Analysis plot that is now visible. It is constructed from time-lapse traces of individual cells using the formula described by Toettcher et al [1].

Note that a single cell cycle analysis is performed for each data source that contains phase information. Time series and growth curves are shown for all datasets with single-cell traces, but phase information is only shown when it is available.

As we shall see, phase information can also be inferred using a computational model of cell cycle progression.

Since the phase information is available for the FUCCI Live-Cell Data, we can overlay phase status onto single cell traces.

Click on the Show phase overlay button to superimpose phase information onto single-cell traces.
Click on the Toggle time series display button next to one of the phase traces to show that phase trace in the Time Series plot.

Section II. Simulating and Perturbing the Cell Cycle

The CCB enables you to simulate cell cycle progression using computational models that describe both the relationships between molecular activities and how each molecule affects progression through each cell cycle phase.

The Map Visualization in the left panel shows a representation of the current model parameters. A model contains expression levels per species, species to phase interactions, and species to species interactions per phase. Each species is represented as a circular node, with size proportional to expression level (as a fold change compared to the initial value specified in the model). The species are replicated for each phase, and links between species and between species and the phase transition lines indicate the interaction between those entities, color-mapped from blue (inhibiting) through grey to red (promoting).

Map visualization

Mouseover any species to highlight all interactions with that species, mouseover any link to show all interactions between those species (or that species and all phases), and mouseover any transition line to show all interactions with any phase.

Sliders below the Map Visualization enable the adjustment of model and simulation parameters. Click any phase in the Map Visualization to active the slider tabs for that phase (and vice versa). You can change models with the Model dropdown above the Map Visualization. Only one model can be loaded at a time. Stars next to the model indicate models belonging to the current workspace. Future versions of the CCB will enable users to compare multiple models and simulation runs at the same time.

Reload the RPE-1 Demo Workspace from the Workspace dropdown menu.
Run a model simulation by clicking the green Run simulation button. A red Cancel simulation button will also appear to enable you to stop the simulation. Any error messages will appear in a red box below the Run simulation button. Simulation progress through each cell, and through each phase per cell, are indicated by two progress bars.

Cell cycle progression is simulated for a number of individual cells, which can be adjusted using the Number of cells slider.

The number of cells equals the number of instances of the Gillespie algorithm to run. With large number of cells or particular parameter values, simulations can take a long time. You can cancel a simulation by clicking the red Cancel simulation button that appears after clicking the Run simulation button. Future versions may include the ability to visualize the progression of running Gillespie algorithm as it progresses through the cell cycle.

After the simulation completes you will see 3 new tracks in the Data Browser panel. Note that the first track contains phase information. This is because simulations keep track of the phase of each cell and can indicate when a given cell transitions between phases.

Note that the simulation also produced a 53BP1 trace and a p21 trace. Recall that 53BP1 was measured in the RPE-1 Cell Line Data and can be directly compared to the simulation. However, p21 is a new species present in the model that is not present in the experimental data. Simulating the expression of p21 could be useful for predicting its influence on cell cycle progression or for future experiments that measure p21 dynamics.
Run the model simulation again by clicking the green Run simulation button.

Note that the simulation output has changed, due to the stochastic nature of the simulation algorithm. The average trace should appear relatively similar, however.

Expression Levels

Let's now change the expression level of one of the species in the model. You can think of this as over-expressing a particular protein. Notice that both species in this model—53BP1 and p21—are shown at '1x' levels, as shown in the Expression levels sliders. Although the absolute levels of each species may differ, these represent the basal levels of each protein. Moving the expression level slider for a species to the right therefore results in an fold increase in expression. Sliding to the left results in a fold reduction. Reducing the fold change far enough to the left results in nearly complete elimination of the species. This action is equivalent to experimentally "knocking down" the species.

Increase the level of DNA damage in the model by increasing the 53BP1 Expression levels slider to 2x the basal level.

Note that when you increase the expression of a species, its size increases in the Map Visualization. You may double-click on the small circle to return to the 1x value.

Increasing the levels of 53BP1 is expected to lengthen the cell cycle, since 53BP1 positively regulates p21, which negatively regulates both the G1/S transition and the G2M/G1 transition.
Click the Run simulation button to simulate the model again and determine the change in cell cycle duration.

Examine the average cell cycle length for this simulation. Its duration should have increased from ~23h to ~29h. Note also that the growth curve for the simulated cell cycle becomes less steep as a result of increasing 53BP1 levels.

Species→Phase Interactions

Other important parameters in the model are the effects of species on the transitions between phases. For example, expression of a strong cyclin dependent kinase inhibitor, such as p21, will slow progress through G1 and G2M. Note that p21 shows a blue inhibitory connection both to the G1/S and G2M/G1 transition in the map diagram.

Click the G2M phase in the Map Visualization. Note that this activates the G2M tab for the Species→phase interactions sliders. Now move the slider for p21→G2M to the right until the inhibitory arrow in the map disappears, eliminating the influence of p21 on the G2M/G1 transition. Simulate the model.

Notice that the cell cycle has shortened to ~21 hours. Specifically, the duration of G2M should have decreased. This may be seen in the simulation output tracks and the Cell Cycle Analysis plot. This effect could represent a weaker G2 checkpoint in the presence of DNA damage.

Species→Species Interactions

Another way to alter the model is to change the way that species interact with each other. These species→species interactions can change during the cell cycle, and are specified for each phase. Let's try eliminating the positive effect of 53BP1 on p21. This change is equivalent to disrupting the DNA damage checkpoint in a cancer cell.

Under Species→species interactions, click the G1 tab. Decrease the 53BP1→p21 slider until you have eliminated the positive arrow from 53BP1 to p21 in the map.

This weakens the interaction between DNA damage and p21 expression. Since p21 expression slows the G1/S transition, weakening the interaction between 53BP1 and p21 increases cell cycle length.

Model Fit

It can be useful to understand how well a given simulation run matches to certain experimental data. To do so, you can calculate the fit between any track from the current simulation, and any track from the currently loaded experimental data. Two fit methods, average difference and correlation, based on the stretched and rescaled average trace for each track, are available.

Load the U20S Demo Workspace from the Workspace dropdown menu.
Stretch the data to a common time axis using the Stretch data button.
Click the Run simulation button.
Show only PCNA-mCherry and PCNA Average traces in the Time Series plot (using the Toggle time series display button next to each average trace).
Rescale the PCNA-mCherry and PCNA tracks with the Rescale traces button.
Compare the time series visually.
Collapse the Simulation parameters, Expression levels, Species→phase interactions, and Species→species interactions panels so that the Model fit panel is more easily visible.
In the Model fit panel, select PCNA from the Simulation dropdown and PCNA-mCherry from the Data dropdown.
Click the Compute button.

The model fit will be shown (color-coded for goodness of fit), based on the average difference between samples in the two average traces for the selected tracks.
Select Correlation from the Method dropdown and click the Compute button.

The model fit based on the correlation between the two average traces will now be shown. Feel free to experiment with different simulation runs, and different simulation and data track comparisons.

Toettcher JE, Loewer A, Ostheimer GJ, Yaffe MB, Tidor B, Lahav G. Distinct mechanisms act in concert to mediate cell cycle arrest. Proc Natl Acad Sci U S A. 2009 106(3):785-90.

Section III: Create your own playground workspace to visually analyze your own data

Click the human icon on the top of the page to create your own guest playground workspace, which will take you to a new page that allows you to create a new guest workspace of your own by uploading your own data and model.
On the new page, fill in the information below as needed to create your own guest playground workspace:
- Input Workspace Name and Workspace Description for your new workspace. These two fields are required in order to identify your workspace on the user interface.
- Click Choose Files button to choose csv datasets to upload from your local disk to the server. You can choose multiple datasets by holding down SHIFT or CTRL key while clicking to select files.
  - The csv datasets being uploaded must conform to the following format in order to load into the browser successfully:
    
    <begin metadata>
    Cell Line, cell_line_value_specific_to_dataset
    Name, name_specific_to_dataset
    Other optional comma-separated attribute-value pair metadata elements, one element pair per line
    <end metadata>
    Cell, comma-separated individual cell names corresponding to each species
    Species, comma-separated species names
    Feature, comma-separated feature names corresponding to each species defined above
    
    time series data follows with each row representing each time step:
    
    time step number, comma-separated values corresponding to feature of each species defined above
    
    Note that the required keywords are marked in the format above denoting they must be present in a valid csv dataset. In addition, if a feature name contains phaseclassification, case insensitive, that feature column will be used to classify time steps with the same number, e.g., 1, 2, or 3, into G1, S, or G2M phase, respectively. Refer to https://cellcycle.renci.org/cell_data_meta/u2os_cdt1_gemenin.csv/ for a sample valid csv dataset.
- Select existing system-provided csv datasets as needed if you want to compare your own data with system-provided data. You can select the first item None or skip this selection box if you don't want to include system-provided datasets in your workspace.
- Edit Dataset Display Name for your selected datasets in the format of cell_data_file_name:cell_data_display_name separated by semicolon for multiple datasets. The default display names use your dataset file names which are populated automatically for you when you upload your own data and/or select system-provided data, but can be manually edited as needed. You can skip this input box if you are satisfied with the auto-populated defaults.
- Edit Dataset Description for your datasets in the format of cell_data_file_name:cell_data_description separated by semicolon for multiple datasets. The default descriptions use your data set file names which are populated automatically for you when you upload your own data and/or select system-provided data, but can be manually edited as needed. You can skip this input box if you are satisfied with the auto-populated defaults.
- Click Choose Files button to choose xml SBML model files to upload from your local disk to the server. You can choose multiple files if needed by holding down SHIFT or CTRL key while clicking to select files.
- You can also create your own SBML model data where each subphase follows over each other by clicking Create your own SBML model data button to trigger a dialog popup as shown below.
  
  You can then input all information in the dialog, and click Create button. You should then see a notification message in green notifying your model data file has been created successfully. A new button Delete SBML model data you just created should also be displayed above the green notification message that allows you to delete the model data you just created if needed and repeat this process to create a new model. Note that currently our model creation code includes three species, namely, PCNA, p53BP1, and p21, with expression levels, species-phase interactions, and species-species interactions over each phase all set at reasonable default values. You can freely change these defaults from the left simulation control panel before starting simulation run in your guest playground workspace after it is created. As a future enhancement, we will allow users to input these information initially when creating their models.
- Select existing system-provided SBML model data as needed if you want to compare your own model with system-provided model or compare system-provided model with your own datasets. You can select the first item None or skip this selection box if you don't want to include system-provided model in your workspace.
- Edit display name for your model in the format of model_data_file_name:model_data_display_name separated by semicolon for multiple models. The default display names use your model file names which are populated automatically for you when you upload your own model and/or select system-provided model, but can be manually edited as needed. You can skip this input box if you are satisfied with the auto-populated defaults.
- Edit description for your model in the format of model_data_file_name:model_data_description separated by semicolon for multiple models. The default description use your model file names which are populated automatically for you when you upload your own data and/or select system-provided data, but can be manually edited as needed. You can skip this input box if you are satisfied with the auto-populated defaults.
- Click Create New guest Workspace button. If there were input errors, the error message would be presented on this page without proceeding to create the requested guest workspace. Otherwise, the requested guest workspace would be created and the user would be directly taken to the guest workspace from which the cell cycles of included data sets are visualized, and the included models can be simulated and compared against the data sets. Note down the URL which should be in the format of https://cellcycle.renci.org/guest/<session_id> where the <session_id> is associated with the user's specific working session for the created guest workspace. The guest workspace including all its data will exist on the server for at least 12 hours, then will be cleaned up nightly from the server, but before the guest workspace is deleted, the same guest workspace URL will work persistently.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly