HCS Tutorial 3

Data:

This data set is from an HCS experiment to characterize how camptothecin affects cell cycle progression.

To follow the steps in this tutorial, you are welcome to request a copy of the trial edition of Panmo and a data set very similar to the one used here.

Assay:

Cells are treated with camptothecin for 24 hours at a gradient of concentrations:

Readout:

Three cellular parameters are monitored:

Channel 1: DNA stain (nuclear stain intensity; DNA content).
Channel 2: Cellular target 1, p53, (nuclear stain intensity). The response of p53 is that its level in nuclei goes up or down.
Channel 3: Cellular target 2, pRb, (nuclear stain intensity). Its response is just like that of p53; its level goes up and down in nuclei.

DNA Content

Since we are looking at the effects of a cell cycle drug, the first thing to look at would be the DNA content, which is our first channel. We are going to use the following procedures to make a trellis graph of histograms of DNA content:

[ Note: The following steps are captured in this 32-second screencast. ]

Invoke Histogram in the primary console:

to get a histogram menu:
Make the following selections:
After clicking the O.K. button in the above menu, we'll get the following menu to specify how to draw this trellis graph:

Panmo remembers the settings of Panel Order, Strips?, Packet Size Bars?, and Abbreviation Threshold used last time. If you are actually running Panmo to follow this tutorial step by step, what you see on your computer might not be the same as this screen shot. Adjust the settings of these 4 parameters to be the same as those shown here; otherwise, you won't get the same trellis graph as this tutorial will get later.
We move Well from the attributes column to the selected attributes column by
- Drag and drop, or
- Click the left mouse button on it and press the S key on the keyboard.
After we do that, this trellis menu will become:

Note that Well is printed on a light yellow background because it is a categorical variable. Numerical variables are printed on a light blue background. Also note that the number of columns and the number of panels are updated to 8 automatically because Well has 8 different categories.
We next adjust the Columns and Rows fields to make the trellis layout the same as the well layout on a microplate:
Click the O.K. button in the above trellis menu to get the following trellis graph of histograms of DNA content:
The default number of bins in a histogram is 20. Let's change the number of bins in each of the histograms in the trellis graph to 50. Click the right mouse button on the trellis graph to pop up the associated right-click menu:

Select Set Histogram Number of Bins... from the right click menu to get the following dialog:

We can drag the slider, type in a number, or press one of the 2 triangles to change the number of bins. Do so to make the dialog look like:

At this time, the trellis graph of histogram of DNA content will look like the following (and you can click the close button to make the above dialog to go away):

It is clear from this trellis graph of DNA content that camptothecin has different effect on cell cycle progression at different concentration. The "2N" and "4N" peaks are quite distinct in the the upper 5 panels. The distribution of DNA content in each panel also spreads out quite a lot. This is because the particular cell line used in this experiment shows a wide range of chromosome numbers.

The 2nd Channel, p53

We'll use a boxplot to visualize the effect of camptothecin on p53. Take the following steps to make a boxplot with a log-scale vertical axis:

[ Note: The following steps are captured in this 17-second screencast. ]

Invoke Boxplot in the primary console:

to get a boxplot menu:
Make the following selections:
Click the O.K. button to get a boxplot:

This boxplot clearly shows that the level of p53 in nuclei jumps up a lot at higher concentration of camptothecin. The median of the H1 well is 4 times bigger than the median of the A1 well.

The 3rd Channel, pRb

Similarly, we can get a boxplot with Well as the X variable and Nucleus Intensity (pRb) in log scale as the Y variable:

This boxplot shows nicely that there is a trend here: As the concentration of camptothecin goes up, the level of pRb goes down.

Point to Ponder

If the expression level or phosphorylation state of a cellular protein changes in different phases of cell cycle, does it make sense not to examine a target protein in terms of the phases of cell cycle?

We'll take cell cycle into account to see how p53 and pRb are affected by camptothecin in the rest of this tutorial.

Dividing Cells into 3 Categories Based on DNA Content

Here we are going to divide all cells in this experiment into 3 categories:

Those in the 2N phase of cell cycle,
Those in the 4N phase of cell cycle, and
Those in the S phase of cell cycle.

We'll do it all graphically by making a histogram of DNA content of all cells and using a paint brush. Here are the steps:

[ Note: The following steps are captured in this 29-second screencast. ]

First, we use the smart Browse command to make a histogram:

Browse will automatically draw an appropriate plot based on the number of and the types of the variables chosen by us.
Clicking the Browse button in the primary console will pop up this menu:
Select DNA Content:
Clicking the O.K. button in the above menu to get a histogram:
Next we are going to divide cells into 3 groups based on their DNA content. We'll do so with Panmo's paint brush. The above histogram is already in paint mode and the color of the paint brush is green. When we move the cursor over this histogram, the cursor will change to a paint brush. To actually paint anything, we have to press and hold down the left mouse button to get a rectangle. The color of the rectangle is the current paint color. We just drag this rectangle around; anything touched by this rectangle will be painted. We paint those cells between the 2N and the 4N peaks green. These green cells are in the S phase of cell cycle. The histogram looks like the following now:
We'll paint cells in the 4N peak red. Before we can do that, we'll have to change the color of our paint brush. Click the right mouse button on the histogram to pop up its right-click menu:

After we invoke Set Paint Brush Color..., we'll get a color palette:

Select the red color and click the O.K. button in the color palette. Now we can proceed to paint cells in the 4N peak red:

What we just demonstrate is essentially equivalent to creating on the fly a new categorical variable with 3 categories
- Blue (cells in the 2N phase of cell cycle)
- Green (cells in the S phase of cell cycle)
- Red (cells in the 4N phase of cell cycle)
For those of you not actually running Panmo to follow this tutorial step by step,here is a screen shot of the computer screen up to this point. Note that this screen shot was done after we hid away Boxplot 3.

Visualizing p53 Responses in Different Phases of Cell Cycle

Now we have divided all cells into 3 groups, each corresponding to one phase of cell cycle. Let's visualize this information on p53 responses, phases of cell cycle, and drug concentration with a trellis graph of boxplots. Because Panmo treats each plot as a viewport into the data space and allows us to retrieve any data visible in a viewport, we'll just use the data in the above histogram to make this trellis graph of boxplots. Note that the above histogram contains exactly the same set of cells as the primary console because there are no missing values in this particular set of data. Here are the steps:

[ Note: The following steps are captured in this 62-second screencast. ]

Click the left mouse button over in the above histogram to enter the data retrieve mode. This is what the above histogram in the data retrieve mode should look like:
Move the cursor over the drawing area (the area with white background in this example) and click the right mouse button to pop up its right-click menu. We are going to invoke Retrieve Displayed Data:
After we invoke Retrieve Displayed Data, the primary console will disappear and a temporary console will show up to contain all the cells in the histogram:
We click Visualization to get a drop-down menu. We are going to invoke Boxplot.
Make the following selections in the ensuing boxplot menu:
After clicking the O.K. button in the above menu, we'll get this trellis display parameter dialog:
We select Well and adjust trellis layout parameters:
After clicking the O.K. button in the above dialog, we get a trellis graph of boxplots:
We can easily get a panel in a trellis graph blown up for detailed inspection. Just put the trellis graph in the trellis display mode:

and left click the mouse button over a panel we are interested in to get a panel plot. For example, the following left panel plot is from left clicking the Well: B1 panel in the above trellis graph and the following right panel plot is from left clicking the Well: H1 panel:

Note that the above 2 plots are constrained to have the same drawing range as that used in each panel of the original trellis graph. This is necessary to make easy, effective comparison between panel plots. If we want to have maximum details, we have to forgo this common drawing range. We can do this by clicking the left mouse button while holding down the Shift key. The following 2 panels plots are generated this way and are from the same panels as the above 2 panel plots:

Visualizing pRb Responses in Different Phases of Cell Cycle

Following similar steps, we can easily get the following trellis graph of boxplots:

Note that the pattern in the H1 (the highest concentration of the cell cycle drug) panel is distinctly different from those in other panels.

Visualizing the Relationship Between p53 and pRb Conditioning on Drug Concentration and Phases of Cell Cycle

Finally, we are going to take drug concentration and phases of cell cycle into account to explore the relationship between p53 and pRb. We do so by making a trellis graph of scatterplots of p53 versus pRb. Here are the steps:

[ Note: The following steps are captured in this 30-second screencast. ]

Invoke Scatterplot from the primary console:

to get a scatterplot menu:
Make the following selections:

Note that in addition to clicking the X and Y toggles, we can press the X and the Y key to switch between specifying the variable plotted along the horizontal axis and the variable plotted along the vertical axis.
After clicking the O.K. button in the above menu, we'll get the following trellis display parameter dialog:
We select Observation color as the first conditioning variable:

Blue cells are in the 2N phase of cell cycle. Green cells are in the S phase of cell cycle. Red cells are in the 4N phase of cell cycle.
We then select Well as the second conditioning variable:
After clicking the O.K. button in the above dialog, we'll get the following trellis graph of scatterplots: