OmicSoft BoxPlot

From Array Suite Wiki

Jump to: navigation, search

Contents

Creating a BoxPlot

Summary

One common way to plot expression data, such as RNA-seq FPKM values or microarray intensity, is as a "Variable View" grouped by metadata (such as tissue, disease state, etc.). These data can then be grouped by metadata, and the range of expression can be summarized with a BoxPlot.

[back to top]

Input Data Requirements

This will work on any -Omic data object, or a Table object containing a series of numeric values.

Step 1: Create Variable View of expression data

The user can add a "Variable" view by right-clicking on the -Omic data object and clicking "Add View", then selecting "Variable View":

Omic AddVariableView Menu.png

In a Variable View, one plot will show the expression for one variable (e.g. gene), plotted for each sample.

Omic AddVariableView DefaultResult.png

[back to top]

Step 2: Group expression data by metadata

Samples can then be grouped by metadata (e.g. tissue):

Omic AddVariableView GroupExpression.png

to show the range of expression by that category:

Omic AddVariableView GroupedResult.png

Step 3a (optional): Color by additional metadata

The individual data points ("symbols") can also be colored by metadata, such as detailed tissue information, and also add "Jitter" to reduce overlap of data points:

Omic AddVariableView ColorByMetaData Window.png

which will allow you see similarities and differences in expression within primary grouping:

Omic AddVariableView ColorByMetaData Result.png

[back to top]

Step 4: Add BoxPlot

Now the user can change the "plotting profile" by clicking Change Profile Gallery:

Omic ChangeProfileGallery Window.png

[back to top]

Classic BoxPlot

The "Classic" BoxPlot will display a box bounded by the 25% and 75% ranked sample, with a line indicating the median, and the "whiskers" will indicate the range to the minimum and maximum value for that group. All values will be plotted unless explicitly filtered out under "Outlier option".

Omic AddVariableView ClassicBoxPlot.png

[back to top]

BoxPlot (R-style)

The "R-Style" BoxPlot will display a box bounded by the 25% and 75% ranked sample, with a line indicating the median, and the "whiskers" will indicate the bounds of 1.5*IQR (Inter-quartile range). Only values outside the whiskers will be plotted.

Omic AddVariableView RStyleBoxPlot.png

[back to top]

Revised BoxPlot (RBoxPlot)

The "Revised" BoxPlot (RBoxPlot) will display a box bounded by the 25% and 75% ranked sample, with a line indicating the median, and the "whiskers" will indicate the bounds of 1.5*IQR (Inter-quartile range). Like the Classic BoxPlot, all values will be plotted.

Omic AddVariableView RevisedBoxPlot.png

[back to top]

(Optional) Creating a Land-style BoxPlot

ArrayLand expressions are by default plotted as "Revised" BoxPlots, but FPKM values are log2-transformed. To emulate the plots you see in Land:

  1. Select RBoxPlot under Change Profile Gallery
  2. Select Flip X/Y Axis under Task | Properties in the View Controller
  3. Select Change Fill Properties under Task | Properties, change opacity to 50%, and change fill color to red.
  4. Select Specify Transformation under Task | Data, and select Log2xPlus0_1

Omic AddVariableView LandBoxPlot.png

[back to top]


Related Articles

EnvelopeLarge2.png

[back to top]