The Violin plot in Partek Genomics Suite is similar to the Profile Trellis plot in that it displays probe(set)/gene intensity values across samples and genes. However, the Violin plot has additional options not shared by the Profile Trellis plot. Here, we will explore one use case for the Violin plot.
For this example, we will use the data set and lists created in the Gene Expression tutorial. We have a list of 23 genes that are differentially regulated in tissue samples from patients with Down syndrome and normal controls. We want to display the mean intensity values for Down syndrome and normal samples for each of the 23 genes on a single plot. To do this, we first need to filter the probe intensities spreadsheet to include only the intensity values for the 23 genes of interest.
With the probe intensities spreadsheet and the gene list open in the Analysis tab, follow these steps to filter the probe intensities spreadsheet.
|
The Filter Columns dialog will open (Figure 2).
|
A black and yellow horizontal bar will appear at the bottom of the spreadsheet. This is the filter indicator showing the proportion of columns (genes/probesets) filtered out (black) and retained (yellow). To continue working with the filtered probeset intensities, we can clone the filtered spreadsheet.
|
The cloned spreadsheet is a temporary file. To ensure we can use it again if we close Partek Genomics Suite, we should save the filtered probe intensities spreadsheet.
)Now we have a spreadsheet containing only the probe intensity values for our 23 genes of interest (Figure 4).
|
We can now invoke the Violin plot. Make sure to have the filtered probe intensities spreadsheet selected (in blue) in the spreadsheet tree as shown (Figure 4).
A Violin Plot tab will open (Figure 5). This plot shows the intensity value ranges of the 23 genes (probe sets) for all samples as violin plots.
|
We can now see the plot properties panel to the left of the violin plot (Figure 6).
|
Although it is called the Violin plot, this visualization can also be used to display box and whisker plots, error bar plots, and gradiant plots. For this example, we will generate box and whisker plots, summarized by Type (Down syndrome and normal), for each gene.
The modified plot shows box and whisker plots, Down syndrome samples in red and normal in blue, for each gene (Figure 7).
|
To improve our view of the gene symbols, we can modify the X-axis legend.
|
|
The box and whisker plots are now centered with a mean of zero and scaled to have a standard deviation of one (Figure 10). Similar to a heat map, this makes it easier to visualize which genes are upregulated and which are downregulated. Here, we can see that most of the 23 genes are expressed more highly in Down symdrome patients.
|
Plots can also be split by categorical variables. We can use this to visualize differential expression of genes between Down syndrom and normal patients in different tissue types.
|
|
Several options will need to be reconfigured before we apply this change.
There should now be a sub-plot for each category, in this case there are four sub-plots, one for each tissue (Figure 13). There are no error bars for several plots because there are not enough samples in those categories.
|
These sub-plots can be displayed all together, or individually.
You can now move through the sub-plots by selecting Next >.
This data can also be displayed as a gradient plot (Figure 14) or error bar plot (Figure 15) by changing the Plot type using the drop-down menu in the Style tab. By default, the shading range in the gradiant plot and the error bars show +/-1 standard deviation from the mean.
|
|
The final option, violin plot, cannot be used to display samples grouped by a categorical variable. To view a violin plot, we must remove the Summarize by selection.
The plot now displays violin plots for each gene showing the distribution of probe intensity values for each tissue in a separate sub-plot (Figure 16).
|
|
