1. A closer look at differential binding
You just completed your first differential binding analysis. Congratulations! Let's look at those results in a bit more detail. A good way to get an impression of what is going on with the data and to get an overview of differential binding results is to look at some plots. In this lesson you will use the plotting functions provided by the *DiffBind* package to create a series of plots summarising the differential binding analysis.
2. Creating MA plots with *DiffBind*
The first plot we'll look at is known as MA plot. It visualises the relationship between change in peak intensity between conditions and average peak intensity. *DiffBind* uses the `dba.plotMA()` function to create MA plots.
3. Visualising differential binding
This type of plot shows the log peak intensity on the x axis and the log fold change on the y axis. You can see an example I've taken from another dataset here. Most of the individual data points have been smoothed into a density cloud to make it easier to see the concentration of points in the high-density areas. Points corresponding to differentially bound peaks are highlighted in pink.
MA plots are useful not only to assess the peak intensity of differentially bound peaks together with the magnitude and direction of the change, they can also highlight larger trends in the data. Typically, most peaks are expected not to exhibit any differential binding. As a result of this, most points in the plot should be grouped around the horizontal line corresponding to no difference between groups.
4. Visualising differential binding
Problems with the data often manifest in a systematic trend away from the horizontal line. In this example, based on the raw data, you can see that the log fold change values show a downward bias that increases with peak intensity. These issues can mostly be addressed with appropriate normalisation of the data. If you find that your MA plot looks problematic, now is a good time to go back and rethink your data normalization procedure.
5. Creating Volcano plots with *DiffBind*
The next plot we'll look at is a volcano plot. *DiffBind* uses the `dba.plotVolcano()` function to create this type of plot.
6. Volcano plots
This plots negative log p-values (or false discovery rates) as a function of log fold change. Peaks with significant evidence for differential binding are highlighted in pink. In this way volcano plots display the significance of the change in peak intensity together with the magnitude of the change, providing a quick summary of differences between groups as well as a way to identify particularly extreme or unusual peaks.
7. Creating Box plots with *DiffBind*
The final plot we'll look at in this video is a box plot. *DiffBind* uses the `dba.plotBox()` function to create these plots.
8. Boxplot
A box plot summarises the distribution of peak intensities within groups and facilitates the comparison between groups. The example you can see here has one box for each of the two groups in the comparison on the left and additional pairs of boxes for differentially bound peaks only. The thick line close to the centre of each bocks indicates the median peak intensity whereas the lower and upper edge of the box correspond to the first and third quartile respectively. More extreme observations are indicated by the whiskers extending from the box and as individual points beyond that. This provides a good indication of the location and spread of the peak intensities in different groups and facilitates an easy visual comparison.
9. Let's practice!
Now it's your turn.