What is a biplot, and how should I interpret data points in a biplot?

OmicsAnalyst
, , , ,
#1

A biplot overlays a score plot and a loadings plot in a single graph. In the biplot view, points are the projected observations (samples); vectors (arrows) are the projected variables. If the data are well-approximated by the first three components, a biplot enables you to visualize high-dimensional data by using a 3D space. You can visualize both the patterns (sample distributions) and the features that driving such patterns.

  • Points that are close to each other in the biplot represent observations with similar values.
  • The cosine of the angle between pairs of vectors indicates correlation between the corresponding variables. Highly correlated variables point in similar directions; uncorrelated variables are nearly perpendicular to each other.
  • The cosine of the angle between a vector and a principal component axis indicates the contribution of the corresponding variable to the principal component: the more parallel to an axis is a vector, the more it contributes only to that PC.

To perform biplot visualization:

  1. Select “Biplot (merged view)” under View options in the top menu bar.

  2. Select omics type, features to be displayed and color of arrows using the dialog

    Screen Shot 2022-07-02 at 4.58.01 PM
    Screen Shot 2022-07-02 at 4.58.01 PM914×798 95.2 KB

  3. Biplot visualization: each arrow represent a feature, top 5 features based on differential expression analysis from each dataset are displayed here.

    biplot2
    biplot2804×425 124 KB

#2

More than 2000 genes and 70 taxa were statistically significant in my lists. I selected ten genes and ten taxa. Only five genes and three taxa were displayed.

What are the criteria for deciding how many arrows show up?

Thank you.

#3

By default, it should be the top significant features. Could you share with me your datasets so I can reproduce the issue. Another way is that you can select “Custom” under Feature Selection option and manually enter the taxa or genes to be plotted.

Guangyan

#4

Hi Guangyan,

Thank you very much again for your kind assistance. I tested your suggestion by manually entering the gene names and hitting “highlight” or “reset” (which one is correct? ) Neither button worked for me.

Here are the files used for my analysis. Please remove them after downloading, as usual. These are unpublished data. I appreciate your understanding very much. :pray:

#5

Hi Guangyan,

Although the “advanced options” did not allow me to manually enter the genes/taxa for arrows, I could manually select features from the right panel and show them as arrows in the plot. This is great!

Thank you.